thermodynamic investigations upon carb
TRANSCRIPT
THERMODYNAMIC INVESTIGATIONS UPON CARB ill Ii
FROM PYRIDYLDIPHENYLMETHANOLS— FREE AND COMPLEXED
By
James Charles Rorvath
A DISSERTATION PRESENTED TO THE CRADLE
COUNCIL OF THE UNIVERSITY OF FLORIDA
IN PARTIAL FULFILLMENT OF THE REQUIREMENTS '
THE DEGREE OF DOCTOR OF PHILOSOPHY
[VERSI'D OJ FLORIDA
1978
This is dedicated to the Spirit of Bombastus, for lie was
blessed with the courage ami the desire to seek the Philosopher's
Stone. In so doing he was magnificently rewarded, for instead
i
•- round the tru th .
This is also dedicated to my Mother and to one Sufferin
trd frora Iowa. For without their inspiration this work would
beei . - np I eted
.
ACKNOWLEDGMENTS
It is not possible to acknowledge by name all who hove con-
tributed to this work. This author is exceedingly fortunate- and
proud to have had the benefit of so much assistance and friendship
and therefore extends his heartfelt appreciation and thanks to
those; knowing that, at least in this way, no one has been omitted.
in
TABLE OF CON - continu
Page
'ENDIX 13 3
1. Specific Gravity of Aqueous HC10, Determined as aFunction of Wt 7, KC1C, . 133
2. Degree of Proton&tion of a Pyridylmethanol (pyLOH)in Aqueous HC10, 134
3. Experiments Conducted upon the Carbeniur Ion Speciesto.md to Exhibit Rearrangement in 70% HC1Q, . . . 136
MKUOGRAFHY . - 140
BIOGRAPHICAL SKETCH 145
TABLE OF CONTENTS
Page
ACKNOWLEDGMENTS .
LIS1 OF TABLES
LIST OF FIGURES .
ABSTRACT
INTRODUCTION
The Carbeniunt Ion
Preliminary Considerations
Groundwork to this Research
1 XPERIMENTAI
Synthesis of Ligands
Synthesis of Complexes
I I
.
Elemental Analyses
Reagents and Solvents
I strumental Analytical Methods
RESU3 TS AND DISCUSSION
Synthetic Gonsiderat ions
Thermodynamic Investigations and Measurements
INCLUSION
x ix
vi
vii
J.
1
2
4
18
18
35
35
38
41
41
43
49
49
60
128
IV
LIST OF TABLES
Table Page
I. Special, Commercially Obtained Reasents - - andSuppliers 42
II- Electronic Spectral Data for the Various CarbeniumIon Species., in 70% HCIC^ at 25° 77
III. Values of H_. in Aqueous HC10. at 25° .... 87
IV. Specific Gravity of Aqueous KC10A
Solutions at 25c
'
. . 91
V. Thermodynamic Stability Data Resulting from the "Deno"Titration of the Various Pyrldylcarbenium Ion ?pec:'es,
in HC10. - ELO at 25° . 95
19VI. F nm.r Cher, ical Shifts (6) for Carbeniura Ion Precursors
in Acetone and 1 K HCIO^, and for Carbenium Ton- in 70%HC10. at 25° 110
4
VI
./ ST 0] FIGU ','
I
.'.' "•
Monopyridyldiphenylmethanols (pyLOH)
Bis (2-pyridyl)phenylmethanols (py^LOH)
Thiazolyldiphenylmethaiiols ....Pyr Ldylphenyl-4-fluorophenylmethanols
Phenolphthalein monopositive ions
Visible spectrum of the carbenium Jon derived f
pyridyl-4-methylphenyl—4—f 1 uorophenylmethanol,HCIOa. v , 20.2, 29.7
'+ max
rom «t-
n 70%
4
4
8
15
72
76
7. Visible spectrum of the carbenium ion derived fromPd(II)<pyLOH)(LN)Cl2 , where pyLOH is 4-pyridyl-4-methyl~phenyl-4-fluorophenylmethanoi, in 70% HCIOa. v~ , 20.6.* / f j n max »
ll.tn
?.. Visible spectrum of the carbenium ion derived fromPd (II) (pyLOH) pC^j where pyLOH is 4-pyridyl-4-methylphenyl4-fluorophenylmethanol, in 70% HCIOa. v , 20.6, 27.7.
^ max
9. Values of '! in aqueous HC10, at 25°R n 4
10. Dilution curve for the titration of 4-pyridyl-4-methyl-phenyl-4—fluorophenylcarbenium ion .....Dilution curve for the titration of [Pd(II) (4-pyL)-(Ljj)Ci9l '", where 4-pyL = 4-pyridyl-4~methylphenyl-4—fluorophenylcarbenium ion ......
ution curve for the titration of [Pd (II) (4-pyL) ~C1„]where 4-pyL = 4-pyridyl-4-methylphe tyl-4-fluorophenyl-carbenium ion ...........Hammett Eo . vs. AG (carbenium ion formation)
pi —Carbenium ion 6 relative to external CFCln vs. AG C of same
L2.
13.
14.
2+
7 6
88
93
93
93
118
118
vxa
LIST OF FIGU U5S - contimied
ire Page
15. Hammett -0-^+ vs . carbenium ion 6 relative to externalCFCI3 . . 118
16. AG° (uncoordinated 4-pyridylcarbenium ions) vs. AG (un-
coordinated 2-pyridylcarbenium ions) . . . . . .121
17. &G° (uncoordinated 4-pyridylcarbenium ions) vs. AG°(bis-complexed 4-pyridylcarbenium ions) . . . . . . .121
vm
Abst act of Dissertation Presented to the Graduate Council
University of Florida in Partial Fulfillment of the Requirementsfor the Degree of Doctor of Philosophy
THERMODYNAMIC INVESTIGATIONS UPON CARBENIUM IONS
DERIVED FROM PYRIDYLDIPHENYLMETHANOLS - FREE AMD COMPLETED
By
JAMES CHARLES HORVATH
March, 1978
Chairman: R. Carl Stoufer
Major Department: Chemistry
The syntheses of a series of 2- and 4-pyridyl-R-phenyl-4~
fluorophenylmethanols arc reported; R may be 4-H, 4-CH , or 4-OCH,,
within each series of alcohols. Palladium(II) complexes of the 4-
pyridyl alcohols have been prepared also wherein either a single
alcohol is Included in the coordination sphere of the metal to yield
the so called "mono"-alcohol complexes, or two identical alcohols are
included to yield the so called "bis"-alcohol complexes. The
corresponding 2-pyridyl alcohol complexes could not be prepared pre-
sumably as a result of steric difficulties. These trityl-type alcohols
(and their complexes) arc precursors to stable arylcarbenium Lor
obtained via dissolution in HC10,-H„0, the Ionizing medium,-a I
19r-\\s "Deno" It
, acidity function titration technique, i nmr
chemical shift measurements, and free energy parameter correlate
have afforded a quantitative evaluation of the stability
the • rious carbenium ion species. These invest; is have also
'
ihed Information which elucidates the Llizing i . ?ct con-
coordinated met a] conl tii Lng moiety to a
coinpj :d carbeai urn ion.
Two particularly noteworthy outgrowths have resulted from these
st idies. First, internally consistent electronic .spectre! interpre-
ons have been obtained which indicate that the stabilizing in-
fluence exerted by the metal center on a given complexed carbenium
ion is reflected by the frequency changes of the carbenium ion ab-
sorption brncls detectable upon coordination of the ion. Secondly,
opriate, linearly interdependent free energy correlations have
given rise to the development of arguments which indicate that al-
though complexation (for the cases considered herein) stabilizes
a carbenium ion relative to the corresponding protonated ion,
coordination does in fact destabilize a pyridylcarbenium ion relative
to tl irotonated free ion species.
INTRODUCTION
The Carl; e. iun; Ton
The imperspicuous nature of the term "carbonium ion" stems from
the fact that this nomenclature classification has been applied to all
types of multivalent carbocations throughout; the chemical literature.
This problem has been pointed out by McManus and Pittman (i) and more
recently by Olah (2). These authors have stated that the term "carben-
Lum ion" is a more systematic reference towards trivalent carbocations
(i.e. , R-C ) which contain a sextet of valence electrons This follows
in that these trivalent carbocations are. valence isoelectronic with
such ions as o -ionium and sulfenium. Here the suffix "-eniura" distin-
guishes these ions from their "-onium" counterparts. Furthermore, this
nomenclature directly establishes a relationship to the species which
results from carbene (:CH„) protonation, namely the carbenium ion
-1-
(Ol-j. Therefore throughout this work the term carbenium ion shall be•J
employed as an explicit reference to a trivalent carbocation xoith a
valence electron sextet. Specifically the cations considered herein
are those which are det Lv>-:: from pyridy 1 dipheny.l.methanols upon disso-
lution of these alcohols in suitable, strongly acidic media.
pL~eliminary Co i is id e r ; 1 1 urns
The preparation of various bis(pyridyl)phenylcarbinoJ : tris-
Cpyi Ldyl)carbinols has been reported by J. P. Wibaut e_t al. (3). These
.:• recognized the direct structural and electronic similarity
of these pyridinecarbinols to triphenylcarbinol (triphenylmethanol)
and therefore investigated their halochromic properties in 100% sulfuric
acid solution. They discovered however, that such solutions exhibited
no color whatsoever. Thus, these pyridinecarbinols were not ionized
in strong acid in a fashion akin to that of triphenylcarbinol. That is
there was little, if any, conversion of the pyridinecarbinols to the
corresponding trityl-type carbeniutn ion.
None the less this disclosure provoked further speculation as to
whether or not such tertiary aromatic alcohols could be converted to
the corresponding carbenium ion. Indeed it seemed to be the. case that
the principal difference in the behavior of these pyridine alcohols, as
compared with their benzene homologues, was attributable to the degr
of positive charge development which would supervene upon their disso-
lution in strongly acidic media. Thus, the concentration of positive
charge produced through base site protonation (of the pyridine ring
nit •: atoms) would be sufficient to prevent carbenium ion formation
to the concomitant development of like charge repulsions. There-
fore Lt seemed reasonable that if these basic sites could be chemically
l ordei to prevent protonation upon treatment with strong acid,
it v then b - feasible to generate the carbenium ion.
Moreover, it was recognized that such investigations upon mono-
Ldyldj ' snylcarbinols were very much relevant to this consideration-
:
i.i, Ls. It was expected that this type of simp] py Ldine alcohol
could be converted to a reasonably stable carbenium ion since this
transformation would be accompanied by the development of less posj L\
char, this possibility was substantiated through the work <>,: Smith
: Holley (4) wherein they report,.! tbu - two structural isomers of
lyldiphenylmethanol ware measurably ionized in concentrated
sulfuric acid solution. Thus, it appeared that carbenium ions derived
from these, particular monobasic alcohols could be employed for stability
investigations up na such ionic species as they are produced in suitable
acidic media; and with respect to the possibility of inhibiting pyridine
ail rogen protonation upon carbeniua ion generation, the binding of these
tic nitrogens through coordinate bond formation to an essentially
neutral metal center seemed to be an appropriate method of general
,,.. Lity. Ih ci Eore investigations upon carbenium ions stabilized..,
by this attendant limitation of positive chai / »uild-up would be per-
i. Also, since pyridine is a ligand which is known to exhibil
n-acid behavior (5:1.17), the potential fc ;asuring the enhancement of
a stability as c: consequence of coordinated atom (ion) "back-donation"
established. That is, if the species bound to the pyridine nitrog
'jessed occupied valence orbitals of appropriate symmetry and energy
to interact with the ir-system of the carbenium ion by creating ir-charac-
-nitrogen bond, the- ion could also be so stabilized.
, if the coordinate bond(s) was sufficiently stable as to remain
Intact upon conversion of such complex compounds to the corresponds
carbenium ion(s), these considerations would be experimentally acces-
sible. (The consequences of o- and tt-1 i Intel ctions between a
coordinated carbenium ion and a prosped al center hove, been
disci r elegantly by Richardson (6:10-11) and arc hei
re) : to iiveni . t ly .
)
Groundwork to this Research
In 1966 Bhattacharyya and Stoufer (7) began work in this area of
research. In accord with standard synthetic methods they prepared
various iiionopyridyldipbenylmethanols (Figure 1) as well as bis (2-pyridyl )
[methanols (Figure 2). Preliminary investigations upon the free
alcohols revealed, as expected, that the monopyridyl derivatives were
converted to che corresponding trityl-type carbenium ion upon treatment
r
(2-pyridyl) (3-pyridyl) (4-pyridyl)
R = -H, -CH , -0CH3
, -N(CH3)2
Fig. 1. Monopyridyldiphenylmethanols (pyLOH)
^foB = -II, -CH-, -0CH-, -N(CH
3)2
Fig. 2. Bis (2-pyridyl) phenylmethanols (py2L0H)
with reagent sulfuric acid (96% l^SO^) . The bis(pyridyl)alcohols, h<
ever, were not ionized by this solvent to any appreciable degree. rlhis
agreed with the results of Wibaut et al. (3). Much of the initial
work was therefore directed towards the utilization of the monopyridyl-
alcohols both as carbenium ion precursors and as heterourcr.-.t io donor
species. In conjunction with this, Bhattacharyya and Stoufer success-
fully prepared a number of palladiura(II) complexes of these monopyridyl-
alcohols by employing them as neutral donor ligands. The materials
obtained were well-characterized as the neutral dichlorobis (alcohol)
complexes of palladium(II) . These complexes were of the general for-
mula Fd(II) (pyLOH)2Cl
2, where pyLOH represents the "ionizable" pyridine
alcohol. These compounds were diamagnetic and square planar as expected
for 4-coordinate complexes of palladium(II) ; (See, for instance, the
discussion concerning the complexes of palladiura(II) by Hartley (5:17-19).)
Subsequent investigations by the.-;e workers upon the carbenium ions
derived from these palladium complexes revealed that dilution of the
ionizing medium (i.e. , the 96% R^SO solution) with water afforded the
reisolation of the intact neutral complex. This experimental find
indicated that the pyridine-metal coordinate bond was reasonably stable
in strongly acidic media. In this way a tangible basis for examining
the stabilities of such coordinated carbenium ions was established.
The results of Bhattacharyya and Stoufer served to demonstrate
that the monopyridyldiphenylmethanols were suitable o-donor ligands as
well as carbenium ion precursors. To continue with this work,
Richardson (6) prepared a series of palladium (II) complexes of the bis-
(2-pyridyl)phanylmethanols which had been synthesized previously b>f
tacharyya and Stoufer. These complexes were of the general formula
J.OiOcl., , and were presumably 4-cootdinate about the me!
th the alcohol functioning as a bidentate ligand. It had been
ted that ss a consequence of binding the nitrogen donor sites
'1 coordination to the metal that these compounds could be cor
verted fu the complexed carbeniutn ion(s). It was discovered, however,
rir.i by employing customary experimental methods this ionization was
not achievable. There was no obvious explanation to account for this
result. Perhaps it was the case that the dissolution of these complexes
in strong acid was accompanied by simultaneous rupture of the metal-
nitrogen coordinate bonds. Since there exists a considerable degree
or steric strain in these 2-pyridyl complexes as a consequence of
spatial crowding between the metal center and the carbinol carbon,
this coordinate bond rupture upon acid treatment was not unlikely.
Richardson carried on with investigations on the stability of car-
beniinn ions derived from the 4-pyridyldiphenylmethanols . He recorded
the electronic spectrum of the free alcohols and of their palladium (II)
complexes in neat trif luoroacetic acid (TEA). These solutions were
highly colored thereby indicating carbenium ion formation with TFA as
solvent. The visible region of the spectrum of these colored solutions
revealed the presence of two intense, broad absorption bands which were
mv to be characteristic of the carbenium ions. An examination of
the visible spectrum of triphenylmethylcarbenium ion produced by dis-
solving triphenylmethanol in trif luoroacetic acid showed the same
:ong absorption bands. Indeed when the electronic spectrum of solu-
of these materials in nonionizing media (e.g., methanol or gla-
cial acel ic acid) was examined in the visible region, these strong
i bands were found to be absent. Ct was also observed durin
these spectral examinations that Lhc positions of these absorption
bancls for a gi ifen carbenium ion species shifted upon going fro th
omplexed protonated carbeniun ion to the corresponding metal-
complexed carbenium ion. This suggested that there was a direct rela-
tionship between the stability of the carbenium ion and the attendant
teration in electronic environment associated with pyridine nitrogen
protonation vs . pyridine nitrogen coordination.
Similar band position shifts were also detectable as the p_ara-
bstituent (R) on the phenyl rings in the alcohols was varied within
the series R = -H, to R = -CH„, to R = -OCH.-. In Lhis instance the
bind shifts were attributed to resonance electronic, interactions between
the R-substituent and the carbenium ion center. It was also found that
the stability of the ion was considerably increased when R = -OCH.,
.
This reflected a substantial capacity of para-OCFU to participate in
favorable conjugational interaction with the ir-system of the trityi-
type ion.
Richardson ultimately attempted to establish a relationship between
carbenium ion development in these 4-pyridyldiphenylr.iethanol systems and
measurable proton ( H) nuclear magnetic resonance (nmr) chemical shifts.
To do this the H nmr spectra of TFA solutions of these alcohols and
their palladium (II) complexes were recorded. Then to fingerprint the
phenyl proton and pyridyl proton absorptions the TFA solution spectra
-.-ere compared with the solution spectra of the unionised compounds.
Also to facilitate the resolution and identification of the various
proton signals, these nmr spectra were subjected to computer simulated
analyses. These investigations however, proved to be unsuccessful
'. the proton absorptions of the unionized compounds could not be
c< rrelated unambiguously with the proton absorptions of the correspond-
edrhenium ions.
As a logical extension of the work, upon the pyridylmethanols Wentz
( ) prepared a series of structurally related alcohols by substituting
a thiazole ring for the pyridine ring. These materials were 2-, and
5-thiazolyldiphenylmethanols (Figure 3). The similarity of these
alcohols to the pyridyl alcohols is apparent. In accord with previous
k the neutral palladium(II) bis-complexes of these alcohols were
pared. These compounds were of the general formula Pd(II) (Th0H) oXo ,
with X representing a coordinated halide ion, either chloride (prin-
cipally) or bromide. Wentz had considered that the replacement, of
R = -H, -CH , -OCH ; R' - -H, -CH.,; R" = -H, -CH
i) 2-Thiazolyldiphenylmethanol ; b) 5-Thiazolyldiphenylmethanol(2-ThOH) (5-ThOH)
Fig. 3. Thiazolyldiphenylmethanols
Ldin with thiazole would add a degree of uniqueness to the antic-
ed chemica] behavior of such heterocyclic bases. That is, since
the .••;:' Ldine and thiazole heterorings arc isoelectronic and contain
essentially Identical nitrogen atoms they should exhibit like donor
characteristics. However, because thiazole also contains a thiophene
type sulfur atom, the thiazole-substituted alcohols should be precursors
to carbenium ions with somewhat different stability than those got
from the analogous pyridine alcohols. Turnbo and coworkers (9) had
indeed demonstrated that the thienyl moiety could enhance the stability
of such carbenium ions relative to the corresponding phenyl-substituted
carbenium ion. They did this by determining the equilibrium constants
(K ) for the ieactioneq
R -t 2 H2
$ R-OH + H+
{1}
in reagent sulfuric acid for a series of structurally equivalent
thienyl and phenyl carbinols. The K values were experimentally
measured by employing the spectrophotometric method of Deno and co-
workers (10) . An ordering of the equilibrium data which were obtained
revealed that a given thienyl-substituted carbenium ion is more stable
than the related phenyl carbenium ion. Thus, these data also estab-
ihed that the sulfur atom in the thienyl nucleus was not protonatod
by the sulfuric acid as this development of additional positive charge
would have induced a net destabilization of these thienyl carbenium
Thus, it was reasonable to expect that the thiazolyldiphenyl-
methanols would be sources of triaryl carbenium ions more stable than
e which were got from the pyridyldiphenylmethanols. Furthermore,
LO
i mization of the thiazolyl alcohols cou] I b lie Lted by L'.ie
I chniq used for .studying the thienyl ions since the pyridine
alcohols . shown to be ionizable in concentrated sulfuric acid.
-
• the stability of the thiazolyl ions could be quantitatively
evaluated provided the acid solvent ionized the parent alcohols to an
extent of one hundred percent (100") • In other words, if the degree of
conversion of alcohol to carbenium ions was measurable in terras of the
i pacity of the acid solvent to produce ionization as a function of
aci-i concentration, th^n any equilibrium pertinent to alcohol-ion inter-
conversion would presumably be monitorable. Undoubtedly it had been
with criteria such as these in mind, that Deno and coworkers (10) were
able i" define an acidity function regarding the ionization of aryl-
methanols in concentrated sulfuric acid — water. The results of Turnbo
Iters C9) proved the suitability of applying Deno's "acidity
Ei " technique towards following alcohol — cjrbenium ion equilibria
in the thienyl systems. Therefore it seemed reasonable to employ this
method for quantitative studies upon thxazolylalcohol — carbenium ion
equilibria, provided the alcohols could be completely ionized at a known
solvent concentration.
Uentz made a very important contribution when he demonstrated that
ly of the thiazolyl alcohols were converted completely to carbenium
ion in reagent perchloric acid (70% HC10,) . This was accomplished by
Lning the visible absorption spectrum of these alcohols as perchloric
acid solutions at various acid concentrations. This simple investiga-
. Lon revealed that for relatively high acid concentrations the carb.
Lved from many of these alcohols exhibited a Beer's law depen-
icej but, as these acid solutions were systematically diluted by the
1]
i o Lsured increments of deionized water, this Beer's law
ience was qo longer maintained. From these observations it was
concluded that in the high acid concentration regions carbenium ion
con Loil was effectually 100%, and the result of adding small quantities
of water was to dilute the concentration of the absorbing species (i.e.,
the carbenium ion). However, as more water was added, the equilibrium
described by equation {1}, (which represents the reconversion of thiazolyl
carbenium ion to thiazolyl alcohol), began to be shifted significantly
tc the right as written. A plot of carbenium ion absorbance vs . weight
percent (wt %) HC10, served to reflect these observations. This plot
over a region of high acid concentrations yielded a straight line for
a completely ionized thiazolyl alcohol. This was the Beer's law portion,
of the plot. But, at an acid concentration particular to the alcohol
under investigation, a marked change in slope was observed. That is,
the plot began to deviate considerably from an extrapolated Beer's law
line. It was at this acidity region where the concentration of the
absorbing species (the carbenium ion) was being diminished not only by
being diluted, but also as a consequence of being reconverted to the
nonabsorbing neutral alcohol precursor. Thus, absorbance fall off was
magnified considerably as the ionizing solvent was made progressively
more dilute.
Thus, by applying appropriate acidity function data (made avail-
able by Deno and coworkers (11) for aqueous perchloric acid) Wentz ;.;as
il Le to measure spectrophotometrically equilibrium constants for the
•oration of carbenium ions resulting from the dissolution of free and
plexed thiazolyldiphenylmethanols in reagent perchloric acid. This
itij .Lion therefore allowed a quantitative ordering of the stabilities
12
',' carbenium ions derivable from these heterocyclic basic compounds.
ends in stability which were established by this study, and the
pertii if generalizations which these trends served to wsrranL, are
aed up accordingly:
(i) Carbenium ions derived from 5-thiazolyl alcohols are more
stable than those got from the corresponding 2-thiazolyl alcohols. This
illustrates th.e inherent destabilizing effect that charge repulsion has
on carbenium ion development. Since the sites of positive charge are
three bonds separated in the 5-thiazoiyl ions, vs. two bonds separated
in the 2-thiazolyl ions, and models of these species indicate a likely
"through-space" charge interaction for the 2-thiazolyl ions, this trend
in stability is certainly expected.
(ii) for a particular alcohol, R-group substitution in the para
position on the phenyl rings, for the series R = -H, -C1I~, -OCH_, results
in an increase in carbenium ion stability. This reflects conjugational
stabilization of positive charge development known for this particular
scries of "R" groups as para substituents in trityl-type carbenium
ions. (See, for instance, the results reported in the papers by Taft
nail McKeever (12), McKinley et al_. (13), and Deno and coworkers (10).)
(iii) For a particular alcohol R'-, or R"-group substitution on
the thiazole ring (see Figure 3), and for R' = -H, -CBL, and for R" =
, --CH , results in an increase in carbenium ion stability as "R" is
reased in mass. This exhibits the greater ability of -CH, compared
with -H to inductively release electron density.
(iv) Carbenium ions derived from the palladium complexes, Pd(IL)-
(ThOH)9X~, are more stable than those got from the corresponding free
ohols. This, at least, illustrates the effect of bindin basic
13
sites Ln the Ligands through coordinate bond formation with an essen-
I
' illy neutral species. This obviously minimizes positive charge
development up;-n treatment with strong acid since ligand protonation
now prevented.
(v) In the instances investigated, carbenium ions derived from
; complexes Pd(II) (ThOH)JBr„, are more stable than those got from
the corresponding chloride ligand complexes. This suggests that a
backbonding mechanism is operative through which the metal center
donates ir-electrcii density into empty TT-orbitals of appropriate sym-
tnetry on the coordinated carbenium ion species. Thus, bromide, which
is expected to be a better ir-donor than chloride, should in turn con-
tribute a net stabilizing effect on the ion via donation of it-electron
density into suitable empty metal orbitals.
Two final investigations of consequence were carried out. The
equilibrium constant for the conversion of triphenylmethanol to tri~
phenylcarbenium iun in perchloric acid was measured. The value ob-
tained was found to be in good agreement with the value which had been
reported for this ionization by Deno and coworkers (11) . This served
further to verify the reliability of the thermodynamic data got for the
generation of thiazolyl carbenium ions. And lastly, the ionization of
U--pyridyJ di (p-tolyl)methanol in perchloric acid was examined. The 4-
pyridyldi'S-tolyl)carbe-Liiuin ion was found to be more stable than the
2-thiazoly3 carbciiun ions but less stable than the 5-Lhiazolyl carbenium
Ions. This result was significant in that it allowed a comparison of
:e hoterorings to be made, as if they were position isomers, with
t to their ability to stabilize trityl-type carbenium ions. More
ortantly, tnis resulc demonstrated the appropriateness of the
14
ctrophot» Lc technique of Deno and coworkers (10) for studying
I hyJ carbenium ions. Therefore carbenium ions derived
a the pyridine alcohols previously investigated by Bhattacharyya and
Stoufer (7), and by Richardson (6), could now be studied quantitatively
and broaden considerably the scope of this work.
The research reported in this dissertation deals principally with
investigations upon free and complexed carbenium ions derived from
pyridyidiphenyJ-iaethanals . These pyridine alcohols and the complexes
thereof were prepared such as to be especially suitable for thermody-
namic stability studies.
The alcohols considered are specifically 2-, and 4-pyridylphenyl-
4- fiunropheny .(.methanols (Figure 4) . The 4-fluorophenyl ring has been
incorporated into the molecular framework of the pyridylmethanols to
19provide a F nmr probe uniquely sensitive to the development of
positive charge upon carbenium ion formation. Considerations for the
19application of F nmr techniques towards stability studies on these
ions were prompted by the unsuccessful H nmr investigations which, for
similar purposes, had been attempted by Richardson (6). The single
fluorine nucleus is particularly suitable for use as a diagnostic nmr
tag in these systems. The principal reasons are the following. First,
h but one such resonating nucleus in the species under investigation,
the spectrum obtained is not complex and is therefore amenable to
straightforward interpretation. Secondly, fluorine in the 4-positLon
on a pheny] ring is known to be highly sensitive to changes in electron
density in the Tr-system of the ring. See, for instance, the papers by
L. (14,15), Dewar and Marchand (16), and Pews, Tsuno, and Taft
15
r-<Cs V
R = -H, -CH -OCH„
a) 2—Pyridylphenyl—4—fluoro— b) 4—Pyridylphenyl—4—fluoro-phenylmethanol (2-pyLOH) pheuylmethanol (4-pyLOH)
Fig. 4. Pyridylphenyl-4-fluorophenylmcthanols
(17), which report that changes in ir-electron density in an aromatic
19system may be precisely correlated with 4-fluorophenyl " F nmr chemical
shifts. These results therefore indicate that the fluorine nucleus
m-cst participate in Tr-bonding interactions with the aromatic ring to
19which it is attached. Thus, F nmr chemical shift data obtained for
a "4-f luorophenyl" fluorine would reflect any changes in TT-electron
density throughout a conjugated system in which this phenyl ring was
incorporated. And so, of primary significance is the relationship which
,ts between the magnitude of the fluorine chemical shift for a
particular pyridyldiphenylcarbenium ion, and the thermodynamic stability
of that Jon. The F nmr studies by Filler (18) and Schuster (19) and
th Lr coworkers upon tris- (p_-f 1 uorophenyl)carbenium ion in different
Li Lng solvents demonstrated the suitability of this experimental
16
al Lon by independently determining K. j_ tor this cation-carbinol
19Libi Luni from F chemical shift data.
This wort also focuses upon the use of the Deuo spectrophotometry
ration technique for quantitative measurements of the stabilities of
carbenium nons derived from free and couiplexed pyridylphenyi-4-rTuoro-
. annuls. The thermodynamic data so obtained are then compared
19with corresponding F nmr chemical shift data via correlation analysis
hods. This data treatment is carried out for the purpose of estab-
lishing interdependent relationships existing between the stability
Information got from each of these types of physical measurement.
In keeping with previous work the neutral bis(alcohol) pal ladiura(TT)
complexes, Pd(II) (pyLOH)„Cl„, were prepared and studied by the physical
methods described above. However, since these materials contain two
moles of "ionizable" alcohol per mole of complex, the degree of positive
charge development upon carbenium ion generation is questionable. This
difficulty had been encountered by Wentz (8) in his investigations upon
the thiazolyl complexed carbenium ions. In an attempt to resolve this
problem complexes of the type Pd (II) (pyLOH) (T.„)C1„ were prepared. In
tht ie new materials, L represents a neutral, nonionizable ligand which
remains coordinated upon carbenium ion generation. Thermodynamic
studies on these new complexes yield information which directly relates
the nature of a singly charged coordinated carbenium ion to the stabi-
lizing influence of the metal center. Thermodynamic data are presented
In which are in respect to the following equilibria:
H+pyI
4+ 2 HO t H
+pyLOH + HgO* {2}
1 /
Cl^PdpyL* + 2 H2
t C12L PdpyLOH + H
3
+{3}
CiUPdCpyL)^ + 4 H2
t Cl?Pd(pyLOH)
2+ 2 H
3
+{4}
Equation {2} pertains to the aqueous titration of an uncoordinated
pyridyldiphenyltnethyl carbenium ion. This equation is written to
emphasize that the pyridine ring remains protonated throughout the
reconversion of carbenium ion to alcohol. This transformation is
associated with a positive charge change of 2+ to 1+. Equation {3}
corresponds to the titrimetric reconversion of a singly charged co-
ordinated carbeniuai ion to the neutral palladium complex precursor.
This process is associated with a charge change of 1+ to 0. Equation
{4} is a composite statement of what actually may be at least two
stepwise processes; initially (perhaps) the reconversion of a species
containing two coordinated carbenium ions to a species with but one
coordinated ion, followed by complete reconversion to the neutral
palladium bis-alcohol complex. This process may therefore be associated
with two full units of positive charge change, viz^. , 2+ to 0. Con-
siderations for purposes of critically evaluating the thermodynamic
daLa obtained from investigations upon these equilibria are accounted.
The apposite conclusions which follow have been presented and are care-
fully discussed.
EXPERIMENTAL
Synthesis of Ligands
The pyr id} Idiphenylmethanols which have been employed as hetero-
aromatic donors (and as carbenium ion precursors) in this research,
were prepared by standard Grignard synthetic methods. This generally
involved the addition of an ether solution of the appropriate 2- or 4-
pyridyl ketone to an ether solution of the required Grignard arylmagne-
iiura halide, follov;ed by acid hydrolysis of the salt-like intermediate
to yield the desired alcohol. Since the necessary ketones were also
e in this laboratory, the synthetic methods for their preparation
have been included. A list of the special, commercially obtained
its employed in these procedures, with names of suppliers, is
provided ia Table I.
The same apparatus and assembly was used in the preparation of
h of the ligands (alcohols) and ketones. All glassware connections
e with standard taper ground joint fittings unless otherwise specified.
All glassware was scrupulously cleaned and dried prior to assembly,
ground joints were carefully lubricated by the application of a
lall quantity of; Dow Corning silicone grease. Rotation of the
connected joints within one another assured the deposition of a unifV
i of lubricant. The assembled apparatus consisted of a 3-necked,
18
19
one-liter round bottom flask equipped with a ground gla ! stirring shaft
in th.? center neck. The stirring shaft was fitted with a Teflon stir
naddLe, and the shaft, was lubricated with the minimum amount of 8117
stirrer lubricant, "Stir-Lube," Ace Glass Co., Vineland, New .Jersey.
Stirring speed was regulated with a rheostat controlled electric stirring
motor. The side necks of the flask were fitted respectively with a
250 ml pressure equalizing addition funnel and a one-liter capacity,
Dewar type condenser charged with dry ice during preparacive runs. The
condenser was attached to the flask with a ball joint connection which
facilitated reaction vessel manipulation required to maintain con-
trolled atmosphere conditions throughout the system, Immediately
following asseml i.y the system was purged with a steady stream of dry
nitrogen gas. The nitrogen environment was maintained until the
hydrolytic step was reached. All syntheses were performed using an-
hydrous diethyl ether as solvent. Magnesium metal turnings used for
Grignard reagent preparation were conveniently activated (unless de-
scribed otherwise) by placing the required quantity of turnings into
the dry reaction flask and stirring them vigorously for a period of 24
— 36 hours at ambient temperature. This procedure reduced the metal
to a finely divided gray-black powder which usually reacted readily with
the appropriate aryl halide to yield the desired Grignard (20)
.
Grignard formation was initiated by gentle warming of the reaction
mixture. If this reaction became too vigorous, cooling the flask with
a cold water bath slowed the reaction to an equable rate. The sub-
sequent addition of reagents to the Grignard (in situ) was done at
reduced temperature by cooling the reaction flask with an insulated
20
bath containing a mixture of chloroform and dry ice. The temperature
of the bath was regulated by the. addition of dry ice as Deeded.
<henyl -2--pyridylketone . Fluorophenylmagnesium bromide was
prepared essencially by the method outlined by McCarty and coworkers
(21). A solution of 63 ml (95 g, 0.54 mol) 4-bromofluorobenzene in
200 ml ether was added dropwise to 13 g (0.53 moi) of activated uiagne-
sium. The mixture was slowly stirred, and the reaction proceeded smooth-
ly as evidenced by the gentle ebullition of ether and the formation of
a brownish sludge. Following the complete addition of the ether —
aryl halide solution the mixture was brought to gentle reflux by
i irming the reaction flask with a "Glas-Col" heating mantle. Grignard
formation was presumed to be complete following reflux for a period of
10 • 12 hours.
The remainder of the procedure paralleled the method of de Jonge
e^ al . (22). The f luorobenzene Grignard solution (above) was cooled
to -35°. A solution of 26 g (0.25 mol) 2-cyanopyridine (picolino-
oltrile) in 200 ml ether was then added dropwise to the Grignard. This
immediately resulted in the separation of a tan-colored solid. The
reaction mixture was stirred continuously during the addition of the
2-cyanopyridine to prevent lumping of the tan solid. After all of the
?.--cyanopyridine had been added the cooling bath was removed, and stirring
continued until the reaction mixture warmed to ambient temperature.
; and contents were then cooled to -30°, and the ketinline
dition compound was hydrolyzed by the careful addition of 50 ml ice
water. This . followed by the addit t 0° of 100 ml. concentrated
I so] i Lng in the formation of a yellow (upper) ether layer
21
and a red brown (lower) aqueous — acid layer. The ether layer was
drawn off and discarded, and the aqueous layer was treated carefully
with concentrated Nil solution until a pH of 6 — 7 was obtained. This
resulted in the separation of copious quantities of a yellowish
precipitate. This material was washed with deionized water and then
shaken with sufficient fresh ether until ail the solid was red if. solved.
The other phase was evaporated to yield 40 g (80%) of the ketone, a
light tan solid which melted at 79 — 81°. The kecone was purified by
vacuum sublimation to yield a white crystalline solid melting at 83 —
84°.
4--Methylphenyl-2-pyridylketone (p_-tolyl-4-pyridylketone) . A hexane
solution of ri-butyilithium (63 ml, 1.6 M) was placed in the reaction
flask and diluted with 250 ml ether. This solution was cooled to
-40°, and an ice-cold solution of 10 ml (17 g, 0.11 mol) 2-bromopyridine
in 100 ml ether was added dropwise with stirring. During the addition
of the bromopyridine the reaction mixture became an orange slurry
which changed gradually to a yellow-green slurry. Following the
complete addition of the bromopyridine, stirring was continued until
the reaction mixture warmed to -30°. The reaction mixture was then
recooled to -45°, and a solution of 13 g (0.11 mol) p_-tolunitrile
(p-nethylbenzonitrile) in 100 ml ether was added dropwise with stirring.
This resulted in the formation cf a yellow slurry. Following the
addition of the nitrile stirring was continued unti] the reaction
mixture warmed to ambient temperature. The reaction mixture was now
recooled to -40° and hydrolyzed by the dropwise addition of 200 ml
2 M HC1. The ether was distilled off, and the reaction mixture was
to 100° and stirred for 1 hour at this temperature to facilitate
decomposition. The aqueous reaction mixture was cooled in
Lee and neutralized by the careful addition of 6 H Nil., resulting in
the separation of a tan solid. The aqueous mixture was then shaken
with sufficient fresh ether to dissolve the solid. The aqueous residue
was discarded , and the ether was evaporated to yield 12 g (61%) of the
cru'.ie ketone. This material was vacuum distilled (0.010 nun, bp 127°)
auc! collected as a light yellow oil which crystallized on. cooling as
yellowish needles. The needles were, dissolved in the minimum amount of
a hot mixture of n-pentane — dichlorome thane (3:1) cud recrystallized
at dry ice temperature as white needles melting at 42 — 43°.
Phenyl-4-pyridylketone (4-benzoylpyridine) . This ketone was prepared
in accord with the method employed for the synthesis of 4-fluoro-
phenyl-2-pyridylketone (p. 20). The Grignard was prepared by the
addition of 50 ml (74 g, 0.47 raol) bromobenzene dissolved in 75 ml
ether to 10 g (0.-'-2 mol) of activated magnesium which was covered with
50 ml ether. Following the addition of the bromobenzene solution
the reaction mixture was refluxed with stirring for 2 hours.
A mixture of 21 g (0.20 mol) 4-cyanopyridine (isonicotinonitrile)
in 2C0 mi ether was refluxed until the nitrile dissolved. This
solution was then added dropwise to the cooled Grignard (-40°). This
:ulted in the immediate formation of a tan solid, and the reaction
ii if was stirred vigorously to prevent lumping. The ketimine
intermediate was cooJed to -55° and hydrolyzed by the addition of 50 ml
I ] saturated aqueous NH.C1. This was followed by the addition
of 100 ml coned HC1 at 0°, resulting in the formal" ion of much rust-
23
tlored solid. The further addition of acid (100 ml 6 M HC1) dis-
' red this solid. The ether layer was separated and discarded.
Adjustment of the pH of the aqueous phase by the careful addition of
coned NH-j resulted in the separation of copious quantities of a yellovz
precipitate. This material was dissolved in the minimum amount of
fresh ether. Evaporation of the ether yielded 34 g (93%) of the ketone,
a well defined crystalline yellow solid y which melted at 68 — 70".
4 --Me thylphenyl-4-pyridylketone (p_-tolyl-4-pyridylketone) . This material
wn^ prepared in the same manner as 4-f luorophenyl-2-pyridylketone (p. 20)
The Grignard was prepared by the addition of 17 ml (24 g, 0.14 mol)
p_-bromotoluene dissolved in 100 ml ether to 3.7 g (0.15 mol) activated
magnesium covered with 50 ml ether. Following the complete addition
of the aryl halide solution the reaction mixture was refluxed for
2 hours ana then cooled to -10°. This resulted in the separation of
a brown precipitate so the Grignard was not cooled further. A
filtered solution of 10 g (0.10 mol) 4-cyanopyridine in 100 ml ether
was added dropwise with stirring resulting in the formation of a large
amount of tan solid. Warming of this mixture to ambient temperature
did not cause the solid to dissolve. The ether phase was drawn off
by aspiration through a coarse frit filtering stick to remove unreacted
4-cyanopyridine. The reaction mixture was recooled to 0° while
"ring, and hydrolysis was effected by the dropwise addition of
50 ml ice-cold saturated aqueous Nil, Br. This was followed by the
addition of 60 ml 2 H HC1, and the mixture was allowed to stand until
unreacted magnesium had dissolved completely. More acid was added
needed to insure that the pH of the aqueous phase was less than 1.
24
,sous phase was now washed twice with 200-ml portions of
I' ct'u.er. The ether washes were discarded, and the aqueous phase
was adji Lo pH 7 by the careful addition of 6 H NH . This resulted
in I h sparation of a considerable amount of white precipitate. This
trial was shaken with sufficient ether to effect dissolution. Evapo-
ration of the ether yielded 14 g (71%) of the yellowish ketone which
melted at 86 - 89°.
4-Hefchoxyphenyl-4-pyridylketone. (This material had been prepared
previously by Bbattacharyya and Stoufer (7) in accord with the method
of LaForge (23). For the sake of completeness its preparation is given
below.)
A solution of 51 ml (74 g, 0.40 mol) p-bromoanisole (l-bromo-4-
methoxybenzene) dissolved in 160 ml ether was added dropwise over a
period of 1 hour at ambient temperature to 9.6 g (0.40 mol) of activated
magnesium. The reaction mixture was stirred vigorously throughout and
refluxed for 1 hour following the addition of the broraoanisole. The
Grignard was then cooled in an ice bath, and to this was added dropwise
a solution of 21 g (0.20 mol) 4-cyanopyridine in A 00 ml ether. The
reaction mixture was stirred constantly throughout. Following the
addition of the 4-cyanopyridine, the reaction mixture was refluxed for
J hour and then cooled in an ice bath to 0°. Hydrolysis was effected
!>, the careful addition of 50 ml ice-cold saturated aqueous NH.C.1. The
r and aqueous layers were then separated, and the aqueous layer
Lee extracted with 100-ml portions of fresh ether. The ether
is were combined with the original ether layer. The ether fraction
extracted thrice with 200-ml portions of 3 M RC1. The aqueous
25
Era< tioi s were pooled and extracted thrice with 100~ml portions of
h Jther. All ether fractions were now discarded, and rhe aqueous
fraction was ooiled for 1 hour to ensure complete katimine decomposi-
tion. The aqueous fraction was cooled and carefully neutralized with
Lce-cold 3 M NaOH resulting in the separation of a yellow precipitate.
This material was filtered, washed with fresh deionized water, and
air dried. The crude ketone was then dissolved in 100 ml hot chloro-
form. This solution was treated while hot with anhydrous MgSO, and
filtered . The volume of the chloroform filtrate was tripled by the
addition of fresh ether and cooled for 1 hour. The reprecipitated
solid was filtered, washed with ice-cold ether, and air dried to
yield 28 g (71%) of the yellowish ketone which melted at 123 - 124°.
?--lyridylphenyl-4-f luorophenylmethanol . Magnesium turnings (8.1 g,
0.33 mol) were placed into the dry reaction flask, and a small crystal
of iodine was added. The flask was carefully heated with a heating
mantle until the iodine just vaporized whereupon heating was discon-
tinued. As the iodine recondensed the magnesium turnings were stirred
briefly to ensure the deposition of a reasonably homogeneous layer
of Iodine onto the surface of the metal. A solution of 32 mi (51 g,
0.29 mol) 4-brcmofluorobenzene in 200 ml ether was added dropwise to
the activated magnesium. The reaction mixture was stirred continuously
as it was warmed co reflux. Reflux was continued for 2 hours following
the addition of the aryl halide, and the reaction mixture was then
cooled to ••60", A solution of 11 g (0.060 mol) phenyl-2-pyridylketone
(2-benzoylpyridine) dissolved in 300 ml ether was added dropwise to
.' Grignard. The cooling bath was removed periodically to
26
5e the freezing out ot" materials from the reaction mixture.
As t tone solution was added the reaction mixture became red-violet
in color. Following the addition of the ketone solution the cool ;
removed, and the contents of the flask were stirred until
a temperature of -10° was attained. During this time the reaction
mixture became dark brown in color. The addition product was hydrolyzed
at -10° by the careful addition of 25 ml ice water resulting in the
formation of a lemon-yellow ether layer and a pink aqueous layer.
The aqueous layer v.as discarded, and the ether layer was extracted
th ice with 200 — ml portions of 3 M HC1. The extracted ether layer-
was discarded, and the aqueous phase was adjusted to pH 7 — 8 by the
careful addition of 6 M NH„. This resulted in the separation of a
yellow-orange solid. The solid — aqueous mixture was shaken with
ficient fresh ether to dissolve all the solid. The aqueous portion
was then discarded, and the ether solution was combined with 3A
ilecular sieves until incipient crystallization was observed. The
sieves ware removed, and the ether evaporated completely to yield 13 g
(77%) of the crude yellow-orange carbinol. This solid was dissolved
in 200 ml hot methanol and treated with 6 g of wood charcoal. This
cure was refluxed 30 minutes and filtered. The hot, yellow methanol
solution was allowed to stand until crystallization of a yellow solid
occurred. The carbinol exhibited a melting range of 79 — 82°.
Anal.: Calcd for C. QH,,N0F: C, 77.40; II, 5.05; N, 5.0?.J.O X.H
Found: C. 77.51; H, 5.10; N, 5.16.
2-E rid 1-4-methylph nyl-4-f luorophenylmethanol. The Grignard was
ictly as that for 2-pyridylphenyl-4-fluoropheny] nol
27
(above) employing 8,2 g (0.33 mol) magnesium turnings and 31 ml (<^9 g,
nol) 4-bromofluorobenzene . The reaction mixture was then cooled
-50°, <md a solution of 10 g (0.051 mo] ) 4-methylphenyl-2-pyr!dyl-
ketone dissolved in 250 ml ether was added dropwise to the stirred
Grignard. This resulted in the formation of a butterscotch-colored
dispersion. Following the addition of the ketone the cooling bath was
r sinoved, and the reaction mixture was stirred until a temperature of
10" was reached. The reaction mixture was recooled to -10° and hydro-
lyzed by the dropwise addition of 200 ml ice-cold saturated aqueous
NH.C1. This resulted initially in the formation of a white slurry
which slowly became a yellowish emulsion. The emulsion was broken
by filtering through glass wool followed by squeezing through coarse
filter paper. The yellow ether layer was then extracted thrice with
100 — ml portions of 6 M HC1. The ether phase was discarded, and the
aqueous portions were combined and treated carefully with 6 M NH.
until pH 7 was attained. This resulted in the separation of a sticky
yellowish oil. The oil was extracted with the minimum volume of fresh
ether , and the aqueous residue was discarded. Evaporation of the
ether again resulted in separation of the oil. Characterization
of the oil (12 g, 82%) revealed it to be the desired carbinol. The
oil was vacuum distilled (0.010 mm, bp range 160 — 170°); but the
collected distillate remained as a light yellow oil after cooling.
1
: Calcd for C H NOF: C, 77.75; H, 5.56; N, 4.77.
Found: C, 77.87: H, 5.56; N, 4.60.
2-PvL'idy.l-4-methoxyphenyl-4-f luorophenylmethanol . A solution of
6.4 ml (9.4 g, 0.050 mol) p-bromoanisole in 25 ml ether was added
28
h stirring to 1.3 g (0.053 mol) activated magnesium covered
b 25 ml ether. The reaction mixture was heated to gentle rei'iux,
1 the formation of Grignard was evidenced by the development of a
grej -brown translucency. Following reflux for a period of 2 hours
gnard formation appeared to be complete. The reaction mixture
was now cooled to -5° with an ice — salt bath, and to this a solution
of 6.0 g (0.030 mol) 4-fluorophenyl-2-pyridylketone dissolved in 120 ml
ether was added dropwise with continuous stirring. During this addition
of ketone a yellow solid settled out. Following the addition of ketone
the cooling bath was removed, and stirring was continued as the reaction
•
; .
: :cure slowly warmed to ambient temperature. This resulted in the
formation of a light tan-colored suspension with traces of reddish-
purple material dispersed throughout. The reaction mixture was sub-
sequently heated to reflux for a period of 1 hour and then cooled.
rhe hulk of the ethereal solution was removed from the reaction flask
by aspiration, through a coarse frit filtering stick. The material
which remained in the flask was washed three times with 50 — ml
portions of fresh ether. The ether washes were removed by aspiration
I combined with the original ether layer. Upon standing a white
semisolid material separated from the ether. The residual reaction
are was now hydrolyzed at 0° by the careful addition of 50 ml
irated aqueous NH.Br, followed by 150 ml 1 M HC1. This resulted
in the separation of a yellow oil. The aqueous phase was adjusted to
pH 7 by the careful addition of 3 M NH„. This produced a milky dis-
lion of the oil. The aqueous phase was then shaken with fresh ether
until the dispersion cleared, and the aqueous layer was drawn off
discarded. The aspirated ether portions (above) were filtered
?-•)
through a medium frit to separate the white semisolid material. ]
ether filtrate was discarded, and the white materia] was hydrolyzi
on the frit by the addition of a few ml deionized water. This pro-
duced more of the yellow oil. The Oil was dissolved :in fresh ether,
and the oil — ether solutions were combined. Evaporation of the
ether yielded 6.2 g (67%) of the oily carbinol. Repeated attempts to
crystallize this material were unsuccessful. An accurate mass for the
molecular ion of the carbinol was determined mass spectrally. Calcd
for C^K^NO^F: 309.1164. Found: 309.1170 (mean of four determina-19 16 I
tions; deviation, ±2 ppm)
.
4-Pyridylphen3^1-4-f luorophenylmethanol . A solution of 12 ml (18 g,
0.10 raol) 4-bromof luorobenzene in 50 ml ether was added dropwise to
2.4 g (0.10 mol) ether-covered activated magnesium. Stirring was con-
tinuous during the addition of the aryl halide, and Grignard formation
ensued upon gentle warming of the reaction flask as evidenced by the
development of a grey-brown dispersion and the ebullition of ether.
After the aryl halide had been added the reaction mixture was stirred
and refluxed for a period of 2 hours. Subsequently, the reaction
mixture was cooled to -5°, and a filtered solution of 11 g (0.60 mol)
phenyl-4-pyridylketone dissolved in the minimum amount of ether (ca.
200 ml) was added dropwise. This resulted in the immediate formation
of a pink solid. Vigorous stirring was maintained to insure uniform
mixing. Stirring was stopped following the addition of ketone, and
the mixture stood at ambient temperature for a period of 12 hours.
The bulk of the ether phase was now drawn off by aspiration through
30
a coarse (Trie filtering stick and discarded. The residual solid \
I twice with fre.>h 50 - ml portions of ether, and the war re
carded. The solid was subsequently recooled to -5° and hydrolyzed
with stirring hy the dropwise addition of ICO ml saturated ice-cold
sous Nil, Br. This was followed by the addition cf 1 M HC1 until a
pH of 5 — 6 was attained. The aqueous mixture was now transferred
.'. large separatory funnel and shaken with 400 ml ether. The aqueous
(lower) layer was tan in color, and the ether (upper) layer was yellow.
A small quantity of semisolid yellow material resided at the interface
of the liquid layers. The aqueous layer was drawn off, and the semi-
solid was combined with the ether layer and together shaken with four
separate 150 — ml portions of 2 M HC1. This resulted in the dissolu-
tion of most of the solid and a translucent ether layer. Evaporation
of the ether yielded a small amount of brown material which was
discarded. All of the aqueous portions were then pooled resulting
in the development of an opaque dispersion. Treatment of the aqueous
layer with 6 M NH- produced initially a clearing of the opaqueness,
and as the pH was raised to 4 — 5 much white solid separated. The
solid was isolated by filtration and the filtrate again treated with
6 M Nil., to bring the pH to 7. This resulted in the separation of
o white solid which was also filtered off. All of the aqueous
filtrate was discarded, and the combined solid samples were air dried
to yield 15 g (92%) of product which exhibited decomposition to a
1 iwnish oil at. 185 — 190°. To convert any hydrochloride salt to free
rbinol the entire amount of white solid was slurried with 100 ml 1 M
After stan ling ("or 1 hour the solid was separated by suction
filtration, i I with 200 ml of deionized water and air dried. The
; ;i Lated material was a finely divided white solid which incited at
L92 — 194° without appreciable discoloration.
m1.: Calcd for ClgH ,NOF: C, 77.40; H, 5.05; N, 5.02.
Found: C, 77.13; H, 5.09; N, 5.00.
An accurate mass for the molecular ion of the carbinol was determined
18 spectrally. Calcd for G' I'NOF: 279.1058. Found: 279.1052' lo 14
(mean of five determinations; deviation, ±2 ppm)
.
4-Pyx Ldyl-4-methylphenyl-4-f luorophenylmethanol . The preparation of
this carbinol was carried out by the method used for the preparation
of 4-pyridylphenyl-4-f luorophenylmethanol (above). The quantities of
materials employed were: 2.9 g (0.12 mol) magnesium; 14 ml (21 g,
0.12 moi) 4-bromof luorobenzene dissolved in 50 ml ethar; and a
filtered solution of 7.4 g (0.038 mol) 4-methylphenyl-4-pyridylketone
dissolved in 125 ml ether.
Following hydrolysis the aqueous phase x^/as adjusted to pH 1 with
1 M HC1 resulting in the separation of 5—10 ml of a brown oil. This
oil was drawn off; the work- up of the oil is given below. The acidic
aqueous phase was twice shaken with 400 — ml portions of fresh ether.
Each shaking resulted in the separation of a small quantity of yellowish
semisolid material. This material and the ether extracts were dis-
carded. The aqueous phase was neutralized by the careful addition of
6 M NH~. This resulted in the separation of a yellowish solid which
was isolated by suction filtration. Characterization of the solid
(4.4 g) revealed that it was the crude carbinol. The brown oil (above)
was stirred with 300 ml 1 M HC1 resulting in tne formation of a brown
creamy emulsion. The emulsion was extracted thrice with 100 — ml
portions of ether. This removed the translucency from t. leous 1
which was now a light yellow solution. All the organic washes were
discarded, and the aqueous layer was neutralized by the careful
addition of 6 M NH>. This resulted in the separation of a yellowish
solid which was filtered, washed with deionized water, and air dried
to yield 1.2 g of solid which melted at 169 - 172°. This material,
combined with the previously isolated solid, afforded a yield of 50%.
Anal.
:
Calcd for CigH16
N0F: C, 77.75; II, 5.56; N, 4.77.
Found: C, 77.60; II, 5.55; N, 4.82.
An accurate mats for the molecular ion of the carbinol was determined
mass spectrally. Calcd for C inH n,.N0F: 293.1215. Found: 29^.1216
19 16
(mean of three determinations; deviation, ±0.3 ppm)
.
4-Pyridyl-4-methoxyphenyl-4-fluorophenylmethanol. The Grignard was
prepared exactly as was that for 4-pyridylphenyl-4-f luorophenyl-
methano.1 (p. 29) using 1.3 g (0.53 mol) magnesium; and 60 ml (9.0 g,
0.51 mol) 4-bromofluorobenzene dissolved in 30 ml ether. To this at
-5° was added in 100 — ml increments a solution of 5.5 g (0.25 mol)
4-methoxyphenyl-4-pyridylketone dissolved in 600 ml echer. As the
ketone solution contacted the reaction mixture a yellow solid formed
and separated. Following the addition of ketone the stirred reaction
mixture was refluxed for 90 minutes and was then set aside and not
disturbed for a period of 12 hours. The reaction mixture was noi; a
yellow creamy dispersion, and little of the ethereal liquid phase
could be drawn off by suction through the glass filtering stick.
Therefore, the reaction mixture was cooled to -5° and hylrolvzed by
33
the dropwise addition of 50 ml of saturated ice-cold aqueous NH.Br,4
followed by the addition of 100 ml 1 M HC1. This resulted in the dis-
persion of a brown oily material in the aqueous phase. The ether layer
was extracted four times with 125 — ml portions of 1 MHC1 and then dis-
carded. The aqueous portions were pooled yielding a yellow-green
opaque mixture. This mixture was adjusted to pH 7 by the careful addi-
tion of 6 M NH~, and upon standing for 2 — 3 hours a quantity of light
tan solid separated. The solid was filtered, air dried, and dissolved
in a refluxing mixture of 100 ml 4:1 ethylacetate — acetone. After
standing 72 hours, this solution was reduced to a volume of ea . 30 ml
by evaporation which resulted in the separation of a white crystalline
solid. The crystals were filtered, washed with a few ml of ice-cold
ether, and ait dried to yield 4.0 g (41%) of the carbinol melting at
181 - 183°.
Anal.: Calcd for C19H16N0
2F: c >
7 3-77; H, 5.21; N, 4.53.
Found: C, 73.75; H, 5.26; N, 4.51.
An accurate mass for the molecular ion of the carbinol was determined
mass spectrally. Calcd for C. nH_ ..NOF : 309.1164. Found: 309.1167
(mean of six determinations; deviation, ±1 ppm)
.
The Purification of Diphenyl-4-pyridylmethane . The commercially obtained
alkane (mp 120 — 125°) was found to be contaminated by trace amounts
of the corresponding diphenyl-4-pyridylcarbinol (from which the alkane
was probably prepared). This was demonstrated by treating a sample of
the "alkane" with 70% HC10, which produced color characteristic of
alcohol ionization. A visible spectrum of this acid solution gave
band positions identical to those got for a similar (known) solution of
d j phenyl-4-pyr idylcarbinol
.
34
I is column (20 cm x 2 cm i.d.) was fitted with a r;Lopcock
which was inserted a plug of glass wool covered wit'.i a I cm
it . The vertically supported column was tilled ca. half
full with reagent hexane, and the stopcock was opened slightly to
the dropwise outflow of solvent. A hexane slurry of freshly
activated 80 — 200 mesh alumina (Brockman Activity I) was poured into
the column, and as the alumina settled on the sand mat the column was
. Fully agitated to insure uniform adsorbent deposition. A 0.5 cm
thick sand mat was added to the top of the alumina layer in the packed
column, and the level of solvent was adjusted to coincide with the
top of the sand mat. A saturated solution of the alkane was prepared
by stirring 2 g of the alkane into 6 ml benzene. This solution was
filtered and carefully placed on the column. Gravity elution was
carried out: by the dropwise percolation of the following solvents:
1) 250 ml 1:1 hexane — benzene; 2) — 5) 500 — ml portions of ether.
Each of the ether fractions 2—4 was evaporated separately yielding ca .
equal quantities of a white solid. A small portion of each of these
samples of solid was treated with 70% HC10, . In each instance the
resulting solution was virtually colorless. These samples of solid
were combined and dissolved in the minimum amount of hot methanol.
Crystallization afforded a yield of 1.0 g of well developed white needles
which melted at 125 — 126°. A solution of these, needles in 70% 1IC10,
; transparent in the visible region of the spectrum.
35
Synthcs [ s of Complexe
s
I. The Preparation of the "bis" Alcohol Complexes of Palladium (If)
,
[Pd<II)(pyLOH)2Cl
2].
Mchlorobis(4-pyridylphenyl--4--f luorophenylmethanol)pallnd ium(II) . Palla-
dium chloride powder (0.16 g, 9.3 x 10 ' moi) was placed in a 250-ml
-3round bottom flask together with 0.10 g (2.4 x 10 " mol) dry lithium
chloride and 100 ml acetone. The mixture was stirred magnetically and
gently refluxed until all solids had dissolved (ca . 24 hours). The
solution which resulted was deep red-brown in color. To this solution
-30.51 g (1.8 x 10 mol) of solid 4-pyridylphenyl-4-f luorophenylmethanol
was added; immediately the red-brown color changed to yellow-orange.
The yellow-orange solution was refluxed for 24 hours while stirring.
-uone was then removed by distillation until a solution volume of
ca . 20 ml was attained. The reaction mixture was filtered through a
m lium frit, transferred to a small beaker, and treated with 5 ml of
deionized water. A yellow crystalline precipitate developed during
otanding for 1 hour. The precipitate was isolated by suction filtration
through a medium frit, washed on the filter with three 10 — ml portions
of fresh deionized water, and oven dried on the filter at 130°. The
solid was then washed from the filter with 50 ml fresh acetone yield-
, ig a yellow solution. This solution was flooded with sufficient
n-pentane to produce permanent cloudiness and was then allowed to
stand until crystal formation occurred. A yield of 0.15 g (20;"') of
well defined yellow needles was obtained. The. product exhibited
d ixkening at >260° and decomposed to a black oil at 295°.
36
. C Led Eor C36
H2g
N2 2
F9PdCl
2: C, 53.75; H, 3.83; N, 3.81.
Found: C, 58.73; H, 4.03; N, 3.70.
0ichlor'jbis(4--pyrio yl-4-niethylphenyl-4-fluorophenylmethanol)-
palladium(II) . The material was prepared in exactly the same fashion
as for the preparation of the bis(4-pyridylphenyl) complex (above)
_3using C.53 g (1.8 x 10 mol) 4-pyridyl-4-methylphenyl-4-fluorophenyl-
thanol. A yield of 0.17 g (22%) of well defined yellow needles was
obtained. The product exhibited darkening at >240° and decomposed
to a black oil at >260°.
Anal.: Calcd for C-JBLgN^FjPdClj: C, 59.74; H, 4.22; N, 3.67.
Found: C, 60.26; H, 4.33; N, 3.49.
Dichlorobis(4-pyridyl-4-methoxyphenyl-4-fluorophenylmethanol)-
palladium(II) . This material was prepared in exactly the same fashion
as for the preparation of the bis (4-pyridylphenyl) complex (above)
_3using 0.56 g (1.8 x 10 " mol) 4-pyridyl-4-methoxyphenyl-4-f luoro-
phenylmethanol. A yield of 0.16 g (20%) of well defined yellow needles
was obtained. The product exhibited darkening at >240° and decomposed
to a black oil at 250°
.
aal. : Calcd for C^U^O^PdCl^ C, 57.34; H, 4.05; N, 3.52.
Found: C, 57.90; II, 4.21; N, 3.38.
'
'J-h lorobis (_2-pyridyl-4-methylphenyl-4-f luorophenylmethanol) palladium (I I)
[This material was not amenable to the thermodynamic investigations
ich were carried out in this work. (See Results and Discussion, p.
97). However, for the sake of completeness, its preparation is given.
It is the only well defined "2-pyridyl" complex which was isolated.]
In actuality this material was obtained as a side product of the syn-
thetic method which had been designed for the preparation of the "salt-
like" complex (Z , PdLCl~) , where Z is a suitable cation, and L is
the pyridylnethanol.
_3Palladium chloride powder (0.28 g, 1.6 x 10 " mol) was placed in
_3a 250-ml round bottom flask together with 0.0/2 g (1.7 x 10 mol)
_3dry lithium chloride, 0.47 g (1.6 x 10 mol) 2-pyridyl-4-methylphenyl-
4-fluorophenylmethanol, and 50 ml acetone. This mixture was stirred
magnetically as it was refluxed for a period of 2 hours resulting in the
dissolution of all solids and the formation of a deep red-brown solution.
The acetone was then removed by distillation yielding some red gummy
material. The gummy semisolid was redissolved by the addition of 10 ml
fresh acetone reproducing the red-brown solution. A heaping micro-
spatula of tetramethylammonium chloride was dissolved in a mixture of
2 ml acetone and 1 ml methanol. This colorless salt solution was added
to the red-brown solution (above) producing no apparent change. The
addition of 2 ml dichloromethane induced the separation of a reddish
oily material which clung to the inner walls of the flask. After standing
overnight the oily material had failed to crystallize and was redissolved
by the further addition of 20 ml fresh acetone. This solution was heated;
following 30 minutes reflux a salmon-colored crystalline solid separated
with the solution phase now being yellow-orange in color. A second
microspatula of tetramethylammonium chloride was added, and the reaction
mixture was returned to reflux for a period of 2 hours. After cooling,
the salmon-colored solid was separated by filtration. (This solid was
later: shown to be tetramethylammonium tetrachloropalladate(II) . ) The
38
Llow-orange acetone filtrate was flooded with deionized water result-
Li in the separation of a yellow crystalline solid. This solid was
red by suci Lou through a medium frit, washed with deionized water.
and air dried to yield 0.52 g (43%) of a material characterized as the
''bis" alcohol complex (Pdl^Cl^ . This material decomposed to a black
Oil above 195°.
tl. : Calcd for C^H^l^O^PdCl^ C, 59.74; H, 4.22; N, 3.67.
Found: C, 59.56; H, 4.39; N, 3.70.
II. The Preparation of the "Mono" Alcohol Complexes of Palladium ('Q)
,
1 Pd (II) (pyLOI-I) OkJCl,] , where L„Tis Diphenyl-4-pyridylraethane. (See
also p. 33) .
The Preparation of Pd(I I) (pyLOH) (LN)C1?, where pyiOH is 4-Pyridylphenyl-
4-fluorophenylmethanol . In a dry environment 0.01 g (5.6 x 10~ mol)
palladium chloride powder was transferred to a 250-m.l round bottcm
flask together with 0.16 g (5.8 x 10+
mol) dried tetca-n-butylamuioniunv
chloride and 100 ml acetone. This mixture was stirred magnetically as
it was refluxed for 72 hours to dissolve all solids, producing a deep
n 1 -brown solution. To this solution was added 0.14 g (5.7 x 10~ mol)
purified diphenyl-4-pyridylmethane (p. 33). As the alkane dissolved
the color of the solution changed from red-brown to red-orange. This
solution was refluxed for 2 hours and cooled. To this was added 0.16 g
(5.7 x 10 mol) 4-pyridylphenyl-4-f luorophenylmethanol; as the alcohol
isolved, the color of the solution changed from red-orange to yellow-
orange. This solution was refluxed for 1 hour after which acetone was
-.1 by distillation until a solution volume of ca. 30 ml was attained,
mixture was now distinctly yellow with incipient; precipitation of a
yellow solid having begun. Sufficient deionized water (ca. .10 ml)
was added until permanent cloudiness was produced. Tne mixture, was
allowed to stand 48 hours to promote crystal growth, and was then filtered
by suction through a tared frit (medium). The collected yellow solid
was washed with deionized water and dried on the frit at 130° for a
period of 1 hour. A yield of 0.35 g (89%) of well defined yellow
needles was obtained. This material darkened above 245° and decomposed
to a black oil above 270°.
Anal. : Calcd for C36
H29N2OFPdCl
2: C, 61.60; H, 4.16; N, 3.99.
Found: C, 61.34; H, 3.93; N, 3.83.
The_Preparation of Pd(II) (pyLOH) (L>;)C1?, where pyLOH is 4-Pyridy3 -4-
1 ..•.'- hylphanyl--4~ fluorophenylmethanol . This complex was prepared in exactly
the same fashion as that for the 4-pyridylphenyl-4-f luorophenylmethanol
complex (above) . The same quantities of materials were employed together
with 0.17 g (5.7 x 10~* mol) 4-pyridyl-4-methylphenyl-4-fluorophenyl-
methanol. A yield of 0.35 g (87%) of well defined yellow needles was
obtained. This material decomposed to a brown-black oil above 245°.
Anal. : Calcd for C37
H31
N2OFPdCl
2: C, 62.07; H, 4.36; N, 3.91.
Found: C, 62.20; H, 4.32; N, 3.77.
ghg..g.r-c.Prtr;ition o £ Pd(II) (pyL0H)(LY)Cl ? , where pyLOH is 4-Pyridvl-4-
'.::xyphenyi-4-fluorophenylmethanol. This complex was prepared in
exactly the same fashion as that for the 4-pyridylphenyl-4-fluorophenyl-
methanol complex (above). The same quantities of materials were employed
40
together with Q.io g (5.7 x 10~ mol) 4-pyridyl-4-methoxyphenyl-4-
phenylmethanol . A yield of 0.35 g (S4%) ol M defined yellow
is obtained. This material darkened above 180° and deco^po .
to a brown oil above 210°.
Anal. : Calcd for C37H31N2 2
FPdCl2
: C, 60.71; F, 4.27; N, 3.83.
Found: C, 60.94; H, 4.46; N, 3.83.
Tetra-ji-butylammonium Trichloro- (4-pyridylphenyl-4-fluorophenylmethanol)
-
palladate(II) . This "salt-like" complex was prepared separately in order
to establish the fact that it was a stable, isolable intermediate. (See
Results and Discussion, p. 58.)
-4Jn a dry environment 0.12 g (4.3 x 10 mol) tetra-n-butylarcmonium
_4chloride, 0.075 g (4.2 x 10 mol) palladium chloride powder, and 100 ml
acetone were placed together in a 250-ml round bottom flask. This
mixture was stirred magnetically as it was refluxed for a period of 24
hours. The reaction mixture now consisted of a deep red-brown solution
;e containing traces of undissolved white and red-brown solids. The
solution phase was carefully decanted into a clean flask, and to this
-4was added 0.12 g (4.2 x 10 mol) 4-py;:idylphenyl-4-fluorophenylmethanol.
As the alcohol dissolved the solution changed color from red-brown to
orange. This mixture was stirred magnetically at ambient temperature
for a period of 1 hour and was then heated to reflux. Acetone was re-
ived by distillation during reflux until a solution volume of ca. 30 ml
is attained. To the hot red-orange acetone solution was added 30 ml
ether, and this solution was allowed to cool without stirring. To the
cooled solution n-pentane was added in small portions until permanent
Liu ;s was attained. Upon stirring for a period of a few hours a
41
all quantity of yellow-orange crystals developed and settled to the
bottom of the flask. The solution phase, which was now slightly yellow,
is treated again with n-pentiine to reinduce cloudiness. After standing
o might the mixture was gravity filtered through a fluted filter , and
the virtually colorless filtrate was discarded. The crystals which
were collected were air dried, examined under a microscope, and found
to he thin, transparent gold-orange sheet-like needles. This material
melted at 150° to a red-brown oil. A yield of 0.29 g (94%, as based
on palladium) was obtained.
Anal. : Calcd for C .H^Q^OFPdCl.,: C, 55.60; H, 6.86; N, 3.81.
Found: C, 55.35; H, 6.87; N, 3.77.
Elemental Analyses
Carbon, hydrogen, and nitrogen elemental analyses for ligands and
complexes were performed either by PCR Incorporated, P. 0. Box 466,
Gainesville, FL, 32601, or by Atlantic Microlab, Incorporated, P. 0.
Box 34306, Atlanta, GA, 30308. No special handling techniques were
required for either the ligands or the complexes.
Reagents and Solvents
The special, commercially obtained reagents which were employed
i i this research are listed in Table I (p. 42). These materials were
used without further purification unless otherwise specified. The
Table I
Special, Commercially Obtained Reagents - and Suppliers
Reagent Supplier
p_-bromoanisole (l-bromo-4-ne thoxybenzene)
4-bromof luorobenzene
2-bromopyridine
2-cyanopyndine (picolino-nitrile)
4-cyanopyridine (isonicotino-nitrile)
diphenyl-4-pyridyIcarbinol
dipbenyl-4-pyridylniethane
phenyl-2-pyridyl ketone(2-benzoylpyridine)
tetra-n-butylammonium cbloride
p_-tolunitrile (_p_-methylbenzo-
nitrile)
Aldrich Chemical Co., Inc,
Milwaukee, WI, 53233
Aldrich
Aldrich
Aldrich
Aldrich
K & K Laboratories, Inc.,Plainview, NY, 11303
Aldrich
Aldrich
Eastman Kodak Co.,
Rochester, NY, 14650
Aldrich
additional reagents and solvents which were employed were readily
available, reagent grade quality materials, and were used without
further purification.
43
Instrumental Analytical Me th
Proton Magnet ic Resonance Spectra . The "I) nmr spectra were obtained
using a Varian Associates Model A-60A nmr spectrometer operating at 60
MHz. The spectra were taken as saturated carbon tetrachloride solutions
using tetramethylsilane (TMS) as internal standard. The spectra were
examined principally with respect to integrated peak intensity ratios
for the purposes of ascertaining sample homogeneity and molecular
composition.
Mass Spectra . Mass spectra were obtained using an AEi MS-30 mass
spectrometer equipped with a DS-30 data system. Solid and oil samples
were introduced into the ionization chamber via direct insertion probe
at 200° and run at an ionizing voltage of 70 eV.
Infrared Spectra . The IR spectra were obtained using a Perkin-Elmer
Model 337-B grating infrared spectrophotometer scanning the region
4000 — 400 cm . Solid samples were intimately ground with oven dried
reagent potassium bromide and run as pressed semimicro discs. Oily
samples were run as follows: A neat potassium bromide disc was pressed
and supported horizontally. The oil was warmed until it became fluid,
and a drop of the oil was added to the surface of the salt disc. This
technique deposited the oil as a uniform thin film on the disc. The
tctrura was recorded as soon as the oil cooled sufficiently to become
v Lscous.
The IR spectra were used to provide evidence of ligand homogeneity
by establishing the absence of a carbonyl stretching vibration (ketone),
44
the presence of a hydroxyl band (alcohol), and to confirm the presence
of both ligands in the "mixed" mono-alcohol complexes (pp. 38-40).
19Fluorine Magnetic Resonance Spectra . The F rurcr spectra were obtained
using a Varian Associates Model XL-100-15 nmr spectrometer operat i
at 94.1 MHz in either continuous wave or Fourier transform pulse mode.
Fourier transform capabilities were provided with a Nicolet TT-100
computer system equipped with a 16 K capacity memory. All spectra were
recorded using external H„0 as a lock signal. Fluorine resonances were
recorded relative to 10% CFClo in acetone and to neat trif luoroacetic
acid as external reference standards. Operation at a sweep width of
5000 llz afforded an uncertainty of ±15 Hz (±0.16 ppm) in the position
19of observed F signals.
Ligsnd spectra were recorded as 0.05 M acetone solutions, as 0.05 M
10% HC10, solutions, and as saturated 70% HC10, solutions. Spectra of
complexes were recorded as saturated acetone solutions and as saturated
70% HC10, solutions. All solutions were filtered through a coarse frit
directly into the nmr sample tube just prior to recording of spectra.
HC10, solution spectra of the complexes were run in large capacity
nmr tubes employing Fourier transform techniques exclusively to facilitate
signal detection for these very dilute samples. All samples were air
cooled in the sample holder during the recording of spectra in order
to maintain ambient temperature.
Visible Spectra and Molar Absorp tivity Coefficien t IX 1 1 crmfna t ion . Vis Lble
spectra were obtained using a Beckman Model DB- ing spectrophotom-
eter equipped with a Sargent Model SR recorder. Molar absorptivity
coefficients (e, Table II, p. 77) were, determined for carbenium ion
species derived from each of the free and complexod alcohols dissolved
in 70% HC.1.0, . Samples of appropriate size (10 to 10 " g) were weighed
to three significant figures using a Cahn Model 1501 Gram Electrobalance
calibrated in the 1 mg range. Oily samples had to be weighed by dif-
ference. This was done by taring a small finely drawn glass whisker
aad then carefully touching the glass whisker to the oil until a very
small bit of the. oil adhered to the whisker. The weight of the oil
was obtained from the combined weight of the oil and whisker. The
weighed samples were stirred in ca. 9 ml of the acid to complete dis-
solution and then made to 10.0 ml with fresh acid. These acid solutions
were scanned vs . the neat acid as blank spanning the region 760 — 320
run to ascertain the position of A for the various carbenium ionmax
species. Each ionic species exhibited two main absorption bands with
the more intense band appearing at lower energy. If the absorption
maxima were "off-scale", the acid solutions were diluted with the
necessary quantity of fresh acid to produce "on-scale" readings within
acceptable sensitivity limitations of the spectrophotometer readout,
viz ., >10% T (<1.0 absorbance units).
Visible Spectra and the "Deno" Titration Technique . The equilibrium
constant (Kg-p datum pertaining to the. thermodynamic stability of each
of the carbenium species which have been considered in this work was
experimentally obtained as follows: a 70% HC10, solution of the alcohol
(ligand) or complex under investigation was prepared and diluted (if
necessary) with fresh acid until an "on-scale" (viz . , 15 — 25% T)
jctrophotometer reading was obtained with the instrument set at *
46
for Chat particular ionic species. It is not necessary to fix the
concentration of this solution. (See Results and Discussion, p. 92.)
I itermined quantity of this solution (ca . 5 g) was weighed to £
i
significant figures into the special cuvette (described below). (By
employing the same sample size for each titration, data treatment was
greatly simplified.) In order to obtain exactly the same weight for
ouch sample, very minute quantities of the parent acid solution could
be transferred to or from (as necessary) the contents of the cuvette
with the tip of a finely drawn glass rod. The acid solution in the
special cuvette was then transferred to the cell compartment of the
spectrophotometer and "read" at X relative to a samole of the neatmax
acid used as blank. The cell compartment was thermostatted at 20 — 25°
by the circulation of tap water. The special cuvette was removed
from the cell compartment, and the parent acid solution of Lhe carbenium
ion was diluted by the addition of a measured increment (ca. 0.02 —
0.0S g) of deionized water from the special burette (described below).
Following each addition of water the acid solution was carefully mixed
in trhe cuvette; the cuvette was returned to the spectrometer, and the
absorbance reread. This procedure was repeated until the absorbance
of the acid solution had fallen off considerably (<0.20 absorbance
units) thereby indicating a reconversion of carbenium ion to alcohol
(or complexed alcohol) precursor in excess of 50%. Data treatment is
considered in Results and Discussion (pp. 83-103).
The specific gravity (p) of the reagent 70% HC10, was determined
before performing the titrations by weighing accurately (5 sig. fig.)
a measured volume of. the acid in a 10 ml volumetric flask which had
> volumetrically calibrated (to 4 sig. fig.) with a weighed sample
47
of distilled water. Prior to calibration the neck of the flask w
heated and drawn to a fine bore of sufficiently large inner diameter to
permit insertion of a Pasteur pipette for liquid transferral. The flask
was then calibrated by marking the drawn neck at a volume dictated by
the weighed sample of water contained in the flask. Specific gravity
data for water at ambient conditions were used to calculate the volume
of the flask at the calibration mark.
Special Cuvette . A 1.00 cm path length quartz cell fitted with a quartz/
Pyrex graded seal stem was obtained from Pyrocell Manufacturing Co.,
Inc., Westwood- NJ, 07675, The stem was shortened so that the cell
fitted conveniently into the sample compartment of the spectrometer.
A standard taper size 13 ground glass neck was added to the top of the
stem to facilitate the direct dropwise addition of water to the acid
solution of the carbenium ion contained in the body of the cell during
titration. A si.de arm of ca. 4 ml capacity was fused to the stem at an
angle of ca. 75° to the cell. Thus the thorough mixing of water with
the acid solution during titration was readily accomplished by rocking
the cell after each addition of water through an angle of 90°. This
particular cell design also eliminated any problems associated with
sample less upon removal of the cell stopper prior to each addition of
titrant (water) since none of the liquid sample was in contact with the
stopper throughout the titration.
la l Burette. A 5.00 ml capacity semimicro burette equipped with an
automatic refilling reservoir and side arm was fitted with a 5 cm length
irgical tubing at the drip tip. A 12 cm length of 6 mm (o.d.)
48
;ry tubing was drawn to ry fine bore pipette tip at one end.
opposite end of the capillary was inserted snugly into the open
of the surgica] tubing. A small screw clamp was affixed to fcl
gical tubing. With the stopcock opened on the burette, the screw
clamp was adjusted until drops of uniform size were discharged at a
nonvenient rate from the drip tip of the capillary. It was found that
nig a titrimetric run (ca. 30 minutes) drops of water could be col-
lected from this burette assembly which differed in weight by not more
than ±0.0002 g for drops averaging 0.0190 to 0.0230 g, provided the
tip of the capillary was wetted prior to drop size calibration. This
obviated the need to weigh the sample in the cell following each addition
of water; i.hat is, it was necessary only to count the number of drops
collected in order to determine the total quantity of added water at
any given time during the titration.
RESULTS AND DISCUSSION
Synthetic Consideration
On the Preparation of Ligands . The pyridine alcohol carbenium ion
precursors employed as ligands in this work were found to be con-
veniently preparable by the Grignard reagent synthetic routes, outlined
in the experimental section. It was discovered, however, that con-
sistently better yields of product were obtained when the Grignard
reagent was prepared from 4-bromofluorobenzene followed by the addition
of the appropriate pyridyl ketone. That is, in attempts to prepare the
identical alcohol from a 4-fluorophenylpyridylketone and the required
phenyl-type Grignard, much poorer yields were got. These results suggest
that 4-bromof luorobenzene Grignard was readily prepared in good yield as
a reactive intermediate and was a sufficiently potent carbanionic reagent
Co attack the carbonyl carbon of the ketone. The difficulties encountered
in the alternate synthetic route leading to acceptable quantities of
product are attributed to the preparation in poor yield of the necessary
Lgnard intermediate. This was particularly obvious in the case for
Lch 4-rae thy 1 phenyl Grignard or 4-methoxyphenyl Grignard were the
. [uired carbanion. Thus, ketones prepared from these Grignards could
usually be obtained only in poor yield. None the less this was not a
»f consequence since the desired material (the ketone) was
49
r>0
Ly separated from the unreacted starting materials. However, the
jparation of alcohol from parent ketone was very difficult, and it was
necessary therefore to convert ketone precursor to alcohol as completely
possible in order to isolate the desired alcohol as a ketone-free
product. Thus, the preparative route for the alcohols employing 4-brorao-
orobenzene Orignard was the better method.
The filter stick filtration technique employed prior to hydrolysis
in the synthetic procedures for the preparation of alcohols facilitated
the removal of unreacted ketone from the reaction mixture. This technique
utilized the fact that the ketone — Grignard addition compound was
sparingly soluble in the ether solvent whereas the ketone itself was
sioderately to readily soluble. Therefore prior to hydrolysis unreacted
ketone which was dissolved in the ether layer could be drawn off by
ion through the filter stick. Repeated washings of the reaction
mixture with fresh ether, followed by filter stick filtration, afforded
essentially complete removal of unreacted ketone.
On the Choice of Palladium(ll) . The most obvious reason for the incorpo-
ration of palladium(TI) as the central metal species into the complexes
considered in this research is to permit a direct and immediate extension
upon the related work of previous investigators. Richardson (6) and
Ventz (8) both pointed out the suitability of palladi.um(II) to such
investigations owing to its low oxidation state and high penultimate
rbita] ocrupancy. These factors would be expected to contribute
L<- metal stabil i ;'.;.'tion of a coordinated carbenium ion via back donation
of metal 4d_ electron density into the iT-fratnework of the carbenium ion
/ided the Ligand donor atom and carbenium ion carbon atom were both
51
members of the llgand -n-system. The relative inertness of palladium(II)
complexes also suggests that such complexes would be amenable to these
types of studies. Finally, the diamagnetic nature of 4 -coordinate
palladium(II) stemming from its square planar complex geometry, makes
it convenient for nmr investigations on the stability of a coordinated
carbenium ion since no paramagnetic contribution to measured chemical
shifts would be observed.
On the Selection and Preparation of Complexes. The results of previous
investigations of bis palladium(II) complexes of ionizable pyridyl and
thiazolyl alcohols demonstrated the suitability of coordinated carbenium
ions derived from these complexes for thermodynamic stability studies
upon such ions. Therefore, the bis palladium(II) complexes were prepared
in order to enlarge the scope of earlier work through similar studies
upon carbenium ions derived from the fluorine-tagged pyridylmethanols
.
The palladium(II) complexes containing but one ionizable ligand
molecule per complex were prepared so that a one-to-one relationship
could be established between a coordinated carbenium ion and the stabi-
lizing effects exerted by the metal-containing moiety. Indeed, the
development of such a synthetic method would in itself be a novel
contribution to that area of preparative coordination chemistry embody-
ing palladium(II) as the central metal species. This follows in that
there are known many "mixed" neutral complexes of palladium(II) of the
type [Pd(II)XYL L„] , where X and Y are anionic groups, and L.. and L,.
-.iro neutral donor ligands. In none of these complexes, however, are
L ligands pyridine homologs; rather they are usually donors which
exhibit particularly strong 77-acid character. Exampjes of those ligands
52
eh are usually found in these complexes are PR„ (R = alky! or aryl)
,
•SR,,, CO, and various alkenes. Actually, work ha? been repotted con-
n Lug the preparation of such mixed complexes wherein pyridine donors
have been ini Led into the coordination sphere as neutral ligands,
but the bulk of this work has focused upon the use of platinum (I I) a;s
central metal species.
In 1936 Mann and Purdie (24) reported the preparation of mixed
complexes of palladium(II) having the general formula [Pd (II) (P.u„P) (am)-
Cio] > where Bu^P is tri-n-butylphosphine, and am is either aniline,
p_-toluidine, or pyridine. These complexes were obtained via the initial
preparation of the bi.nuclear chloride-bridged trans bis(Bu,P) complex
(Pd„(Bu,P)-Cl, ] , followed by cleavage of the chloride bridges with the
required molar ratio of the amine of choice. However, an examination
or the information cited in the experimental section of this article
revealed that accounts are given only for the preparation of the complexes
which incorporated aniline and p_-toluidine as the nitrogen donors.
Investigations by Chatt and Venanzi (2b) in 1957 upon similar complexes
of p'illadium(II) again served to demonstrate that the neutral chloride-
bridged, binuclear compounds could be converted to the corresponding
mononuclear complexes via rupture of the bridging bonds of the halide
j.oxin with tripheuylphosphine. The parent bridged materials had heen
preparable with di-n-pentylamine or piperidine as the nitrogen donor
ligands Ln trans positions; but these workers reported that they were
able to obtain the related bridged compounds using pyridine or
pyridine-containing ligands. Obviously, therefore, neutral mononuclear
Le 5 with pyridine in the coordination sphere had not been iso] I
during these investigations. In 1969 Chart and Hingos (26) repor
the successful preparation of neutral, square planar, mixed, mononuclear
pli xes of palladium (11) which had pyridine included within the
coordination sphere of the metal. Again the synthetic route to these
:ed mononuclear materials relied upon the cleavage of bridging bonds
in binuclear precursors wherein the bridging species were either
chloride ions or £-toluonesulf inate ions. In 1975 Boschi and coworkers
(27) succeeded in preparing some neutral, square planar, mixed, mononuclear
complexes of palladium (II) with coordinated pyridine. They also employed
a chloride-bridged binuclear precursor [PdJL^Cl, ] wherein the neutral
L groups were aromatic isonitriles. Treatment of the bridged compound
with the required molar ratio of pyridine afforded the isolation, in
good yield, of the corresponding trans mononuclear complex. The results
of both of these studies (viz. , Chatt and Mingos, and Boschi and co-
workers) therefore indicated that a general route towards the synthesis
of mixed mononuclear complexes of palladium(II) which included pyridLne-
fcype ligands required initially the preparation of a suitable halide-
bridged binuclear complex, followed by the rupture of the bridging
bonds with the selected pyridine donor(s).
So, in order to obtain the desired pyridine-containing, mixed
mononuclear complexes in this research, it was first attempted to
prepare the binuclear chloride-bridged dimeric materials [Pd^L^Cl,]
with the L groups as the pyridine methanols. This method depended upon
the direct combination of the mononuclear bis-alcohol complexes [PdL„Cl„]
2-with the complex anion [PdCl, j" on a 1:1 mole basis with respect to
palladium. The bridged dimer, however, could not be obtained by this
method. Therefore, the chloride bridged complex [FdyLJZl.] with tri-
aylphosphine (Ph,,P) ligands incorporated as the neutral L groups was
54
tared accoi lii,
to the method outlined by Chatt and Venanzi (25).
The Ph.^P was employed because of its structural similarity to the pyr-
j.d) sthanols. Thi? bridged material, which was a red-brown solid,
slurried in refluxing acetone and treated with 4-pyridyl-4-f luoro-
lylmethanol (4-pyLOH) on a 1:1 r.ole basis with respect to palladium.
'••; reflux was continued (ca. 3 hours) the bridged material dissolved
and a yellow solution resulted. When this solution was saturated with
pentane a pale yellow crystalline solid settled out. Characterization
of rhis solid indicated that the desired mixed mononuclear complex
tPd(II)(Ph3P)(4-pyLOH)Cl
2] had been obtained. This result suggested
that the problem of preparing the mononuclear mixed complexes containing
the pyridine alcohols had been solved. However, when this material was
treated with 70.'" HCIO^ for the purpose of carrying out stability in-
vestigations upon the "coordinated" carbenium ion it was discovered that
the coordinate bond between the ionized pyridine alcohol and the palladium
metal center was quickly ruptured ( ca . 5 minutes or less) . It was pre-
sumed that this bond breaking was a consequence of the effective trans
labilizing effect exerted by the strong ir-acid ligand, triphenylphosphine.
This result therefore dictated the necessity to prepare the mixed mono-
nuclear complexes with a neutral, nonionizable counter ligand, which
would not exhibit particularly strong Tr-acid behavior in these complexes.
The selection of 4-pyridyldiphenylnethane as an appropriate counter
1 Lgand was clearly a good choice. This is attributable to the structural
parability of the alkane to the pyridine alcohols, and to the antic-
I ed similarity in chemical behavior of the alkane to pyridine it-
Li . Thus, the problem at hand was the development of a synthetic
procedure through which a molecule of the alcohol as well as a molecule
JJ
of the alkane could be. systematically introduced into the coordination
of the metal in the mononuclear complex. The following con-
siderations were pertinent: 1) The inability to prepare the chloride-
bridged binuclear complexes containing pyridine donors in trans positions
IPd« (py'LGH)„Cl, ] (vide supra ) ruled out the use of such a material as
a precursor. The bridging bonds in this dimer would presumably have
been susceptible to attack by the "second" pyridine donor thereby
yielding the mixed pyridine mononuclear complex. 2) In view of the
anticipated similarity in donor character of the pyridine alkane to the
pyridine alcohols there appeared to be no method by ;^hich these two
pyridine Uganda could be added directly to the palladium metal center
in the required stoichiometric ratio.
The synthetic finds of Goodfellow, Goggin, and Duddell (28) provided
reasonable prospects for the preparation of the desired complexes. These
workers recognized that there are many complex anions of the type
[Ft (II)LC13]~ (L - C
2EA> C0 » N0 > ^v or pyridine; references for the
preparation of this complex anion with these respective ligands are
cited in this article) w7hich are well known. They discovered that such
complex anions were also preparable with L as PR.,, SR„ , or AsB.~ (R =
alkyl or aryl), and that a similar series of complexes (excluding
pyridine) was preparable as well with palladium(II) . This was the first
general account given for the successful preparation of such types of
ilex anions of palladium(II) . The usual synthetic route which these
workers employed was characterized by refluxing the chloride-bridged,
binuclear material [M?L^C1, ] , II equals Pt(II) or Pd(II), with the re-
quired stoichiometric quantity of tetra-n-propylammonium chloride in an
rt organic solvent such as dichloromethane. The complex anion which
56
was formed upon rupture of the chloride bridges was apparently Lized
in solution by the large tecraalkylammonium counterion. The complex
anion was found to be isolable as the tetraalkylaronionium salt via treat-
ment of the d.Lohloromethane reaction mixture, with excess ether which
resulted in crystallization of the desired product. The salient aspect
this preparative method is that it permitted the controlled inclusion
of a particular neutral ligand into the coordination sphere of the metal
acorn. However, as previously indicated, this particular method was not
directly applicable to the situation involving the pyridine donors
since the necessary bridged precursors with trans pyridine donors had
not been preparable. Nevertheless, further considerations paralleling
this synthetic approach were certainly warranted in that, this technique
served to reinforce the possibility of being able to investigate stabil-
ity relationships between the metal center and a singly charged co-
ordinated pyridine carbenium ion provided [Pd(II)LClo] complex anions
of the pyridine alcohols could be prepared.
Goodfellow, Goggin, and Duddell had also reported the preparation
of the anionic complexes [Pd(II) (C2H,)C1
3]~, and [Pd(II) (C0)C1_]"*, by
2-reacting the binuclear anion [Pd
?Cl.] with the required molar quantity
of the neutral ligand in the presence of tetra-n-butylammonium ion
i tng cis-l,2-dichloroethylene as solvent. Again treatment of the
reaction mixture with excess ether induced the separation of the desired
product as a crystalline solid. Infrared spectroscopic investigations
bj Adams and coworkers (29) also served to indicate the potential use-
2-fulness of- the binuclear anion [Pd
?Cl,] as a precursor to mononuclear
1
aces of palladium(II) of the type [Pd(II)LCl_] by demonstrating
57
that the force constant for the terminal metal-halide stretching
vibration was greater than that for the bridging metal-halide vibra-
2-tion. Thus, in would be expected that the dimeric anion [Pd Cl,l
/. b
would be attacked at the bridging positions by incoming ligands.
It appeared, Therefore, to be reasonable to treat polychloropalla-
date(II) anions with the pyridine alcohols in an appropriate solvent in
the presence of t etr aalky1ammonium cations with the anticipation of sus-
taining the stability of the [Pd(II) (pyLOH)Cl.,] complex anions. In
conjunction with this, palladium chloride powder was stirred in re-
fluxing acetone with tetramethylammonium chloride (tMACl) on a 1:2
mole basis. A red-orange solution initially resulted as the solids
began to dissolve, but as reflux was continued the color of the solution
disappeared and a salmon-colored solid separated. This solid was slur-
ried with ^-pyridylphenyt-4-fluorophenylmethanol in refluxing solvent but
no additional change occurred. Nevertheless, the incipient formation
of the red-orange solution indicated the presence of a solution-stable
chloroanion of p3lladium(II) . Further investigations indicated that
2-the solution-stable species was [Pd Cl,V and that the salmon-colored
i. b
solid was the acetone-insoluble salt [ (tMA) „PdCl, ] . Thus, the nature
o F the palladium(II) polychloroanion was dependent upon the concentra-
tion of the ammonium salt. Studies by Henry and Marks (30) upon
glacial acetic acid solutions of palladium(II) in the presence of
various alkali metal chlorides served to indicate that with readily
2_soluble chlorides such as LiCl, [PdCl,] was the commonly encountered
palladium(II) anion, whereas with moderately soluble chlorides such
2-as NaCl, the binuclear species [Pd„Cl,] was got. Therefore, it
•eared that regulation of the tetraalkylammonium chloride concentration
53
orded a method by which the nature of the palladium (I I) polychloro-
aion could be controlled. The combination of palladium chloride
der with tMACl (1:1) in refluxing acetone did in fact yield complete
dissolution of all solids and a stable red -orange solution. Treatment
of this solution with a typical 4-pyridylmethauol (1:1 with respect
to palladium) again resulted in the separation of a salmon-colored
precipitate [ (i.MA) 9PdCl, ] and a yellow solution. Workup of the yellow
solution yielded the bis-alcohol complex [Pd(II) (pyLOH)„Cl„] . Thus,
the tMA cation did not appear to be capable of stabilizing the desired
anionic mono-alcohol complex regardless of the nature of the polychloro-
J.adate(II) precursor.
Palladium chloride powder was then combined with tetra-n-butyl-
lium chloride (.1:1) in refluxing acetone again resulting in the
d Lssolution of all solids and a stable red-orange solution. Treatment
of this solution with a typical 4-pyridine methanol (1:1 with respect
to palladium) induced an immediate color change in the solution from
red-orange to yellow-orange with the separation of no solids. Workup
of this solution (see Experimental p. 40) revealed that the desired
anionic complex [Pd (II) (4-pyL0H)Cl_ ] had been obtained. Further
studies showed that this anionic complex was readily converted to the
Lred mixed, neutral, mononuclear complex by treatment (1:1) with
the 4-pyridyl alkane in acetone solution (see Experimental pp. 38-40).
(Note: During the course of this research all synthetic procedures
Involving palladium (II) were run using acetone as solvent. There are
lard methods for the preparation of complexes of palladium(II)
l.oying alcohol (usually methanol or ethanol) as solvent, but in this
' it was discovered! that in the presence of alcohol palladium (II)
59
.; frequently reduced to palladium black. No similar difficulty was
encountered with acetone.)
he Suitability of 4-Pyridyldiphenylme thane as a Counte clif;and in the
Mixed Complexes . As reported previously (see Experimental p. 33) the
commercially obtained alkane was found to be contaminated with trace
quantities of the corresponding 4-pyridyl alcohol. To be sure that
the alkane was not converted to the alcohol (carbenium ion) via oxida-
tion in 70% HC10, . a solution of the purified alkane in the acid was
srirred at ambient temperature in the open environment of the laboratory.
After stirring for a period of 2 hours this solution was scanned in the
visible region of the spectrum and was found to be transparent. There-
fore no complications were expected to arise during thermodynamic
studies upon complexes which contained the alkane since all such mea-
s cments were made within a 2 hour time span.
The similarity in donor behavior of the alkane to the 4-pyridyl
alcohols was demonstrated by the preparation of the anionic complexes
p.VLCl^] (in situ) with either the alkane or the alcohol, followed by
conversion to the mixed complex. Thus, the mixed complexes were
preparable independent of the order of addition of the respective pyr-
idine donors.
Finally, a small sample of mixed complex (any) was triturated in
70% HCIO^, for a period of ca. 1 hour. The acid was removed by filtra-
I ic n and the lisidue was washed with deionized water and dried. An IR
scan of the residue revealed it to be of the same constituency as the
original mixed complex. This indicated that both pyridine ligands re-
ned coord in.-' ted during thermodynamic stability investigations and
GO
;ain '
t rated the suitability of 4-pyridyldiphenylmethane as an
iropriate counterligand. Mien the complex [Pd(II) (Ph^P) (4-pyL0H)Cl? ]
had been treated in a similar fashion (i.e. , triturated in 70% HC10.)
i; was found that the pyridine ligand (carbenium ion) was discharged
from the complex (p. 54).
Concerning Carbenium Ion Salts . Various synthetic routes are available
for the preparation of stable salts of trityl-type carbenium ions (see,
for instance, the methods given in the papers by Sharp and Sheppard
(31), by Daube.n et al. (32), or by Olah et a^. (33)). It may certainly
prove to be interesting and profitable to investigate the stability of
the free and complexed carbenium ions which have been dealt with in this
work as discrete salt-like species in aprotic, nonreactive solvents.
Current considerations however, have exclusively involved the use of
70^ HCIO, as an ionizing medium in order to provide a direct extension
to previous work upon trityl-type carbenium ions derived from the
I dyldiphenylmethanol s
.
Thermodynamic Investigations and Measurements
I; ctronic Spect ra and Carbenium Ton Constitution and Structure . Con-
siderable work has been done on the electronic spectra of trityl-type
carbenium ions in the 200 — 750 r.m (50.0 — 13.3 kK) spectral region.
The majority of this work, however, ha.; been focusod upon the transitions
' by these ions in a rather limited portion of this region,
^00 — 550 nm (i'i.3 — 18.2 kK) , because the electronic bands found
61
h re ire those which are characteristic solely cl the carbenium ion.
higher energy transitions (50.0 — 33.3 kK) exhibited by these ion;-,
nre normally present in the spectrum of the precursor molecules (i.e . ,
tertiary alcohols) and are characteristic of the electronic absorptions
of the isolated conjugated systems which are bound to the carbinol
carbon in the unionized alcohol. Previous electronic absorption studies
on these types of ions (e.g., Richardson (6) and Wentz (8)) including
similar studies performed during the course of this work have demon-
strated that the higher energy (ultraviolet) spectral region of these
alcohols remains virtually unchanged for a given alcohol independent
of the nature of the solvent. Thus, the significant electronic changes
which occur in these systems upon carbenium ion generation are not
directly reflected by these higher energy transitions. The dramatic
changes which do take place in the electronic spectrum of these alcohols
upon ion formation are illustrated by the development of two intense
/_, 5 _] _i(r ca. 10 — 10 liter mol " cm ), broad absorption bands ordinarily'max —
appearing between 300 and 550 rim. This spectral region is transparent
for the alcohols dissolved in a nonionizing solvent, e.g., acetone,
alcohol, or 1 M HC10,. The extreme intensities of these bands are ex-
pected for strongly allowed n +- tr charge transfer type transitions.
As pointed out by Dunn (34) based upon considerations of the classical
theoretical work of Mulliken (35) on electronic spectroscopy, the
.intensity of a charge transfer band is expectably large since the
charge transfer phenomenon occurs over at least one interatomic dis-
tance in the absorbing species. Thus, the radius vector (r) of the
transition is relatively large. Since the magnitude of the transit ton
tent integral is directly proportional to r, and in turn directly
62
p oportional Lo the oscillator strength (f) of the transition, f must
also be large. Hence, the intensity of the transition is considerable.
rhe general positions of these trityl-ion charge transfer bands have
.. rationalized by considering that alcohol ionization is accompanied
bj the conversion of the system from a quasi even-alternant benzene
hydrocarbon to an odd-alternant, fully conjugated benzene hydrocarbon.
According to various workers (see, for instance, Deno et al. (36))
based on simple LCAO MO calculations this transformation introduces a
zero energy (nonbonding) iT-symmetry orbital into the molecular orbital
scheme of the previously unionized alcohol intermediate between the
highest energy filled vr-orbitals and the lowest energy unfilled ir*-
crbitals. Thus, the energy of the longest wavelength (lowest energy)
electronic transition observed for these ionic species should be on the
order of half the energy of the longest wavelength transition exhibited
by benzene, the model compound. Since the wavelength of this transition
for benzene is 256 nm it is expected that the longest wavelength electron-
ic transition of the trityl-type ions would appear in the vicinity of
512 run. As pointed out by Richardson (6:153) the rather inexact nature
of this treatment is revealed by the fact that the longest wavelength
electronic absorption exhibited by triphenylcarbenium ion is 431 nm
(in 96% H-SO^) which is at considerably shorter wavelength than that
predicted from the benzene model.
An eclectic account of previous investigations upon the electronic
nature of arylcarbenium ions reveals that extensive considerations have
been made, but that these considerations are not without certain per-
:ing aspects. In 1932 Schoepfle and Ryan (37) reported that the two
aces, triphenylchloromethane and methyldiphenylchloromethaue, yield
iV3
e ientially the same visible spectrum when dissolved in dichloroethylene
in the presence of stannic chloride. This prompted Newman and Deno (38)
to conclude that this was evidence indicating that in triarylcarbenium
ions no more than two (and perhaps just one) of the aryl rings could
simultaneously participate in resonance interactions with the carbenium
ion center (the exocyclic carbon atom). Completely synchronous resonance
stabilization of a triarylcarbenium ion involving all of the aryl rings
would of course require an all-planar molecular ion configuration of D_
symmetry. Lewis e_t al. (39) had already reported as a consequence of
studies on crystal violet ion (tr is- (dimethyl-p_-aminophenyl) -methyl ion),
that the all-planar configuration of the ion is not possible owing to
sterie interactions between the ortho hydrogen atoms on the phenyl rings.
These workers had speculated on the existence of two isomers of the ion.
having structures akin to a symmetric and an asymmetric propeller wherein
the phenyl rings were the blades of the propeller. The presence of two
intense bands in the electronic spectrum of crystal violet ion supported
the proposal that each of the two isomeric forms of the ion was a distinct
chromophoric system. Additional evidence cited by Newman and Deno which
suggested the structural uniqueness of trityl-type ions was the following.
Tri-o-toiylcarbenium ion was reported to be as stable as tri-p_-tolyl-
carbenium ion and to exhibit essentially the same electronic spectrum.
Tliis x^as an unexpected result owing to the considerably greater degree
of inhibition towards ring resonance stabilization of the ion anticipated
for the tri-o-tolyl ion as a consequence of sterie interaction between
the o_-raethyl groups. Also, van ' t Hoff i-factor data on solutions of tri-
p-dimethylaminophenylcarbinol in 100% H„S0, indicated that even in this
Jngly acidic medium one of the p_-amino groups was not protonated.
64
This suggested that only one of the rings was involved substantially in
tonance stabilization with the cation center. Other investigations
by Newman rid Deno on the electronic spectra of various carbenium ions
revealed that the observed band positions were sensitive to changes in
phenyl ring substitution. Attempts to rationalize these band shifts
>.rc;.iiscd mainly on resonance considerations were inconclusive. Further
attempts to rationalize the observed differences in band intensity and
position in the electronic spectra of various series of related aryl-
carberiium ions by Deno, Jaruzelski, and Schriesheim (40), met with
limited success. These workers discovered a systematic spectral trend
characterized by an increase in A as well as in band intensity re-max 3
suiting from singular substitution of any of the groups, -N(CH«) , -NH,
-0CH_ or -CI, into the para position on one of the phenyl rings in
triphenylcarbenium ion. The trend appeared to coincide with expecta-
tions contingent upon simple extension of the 7T-electron donating, con-
jugated system of the substituted cation relative to the triphenyl ion.
Ho'/ever. successive para substitution of these groups on subsequent
phenyl rings in a given ion resulted in discontinuous shifts in v andr J & b max
in molar absorptivity. In 1954, Branch and Walba (41) studied the
electronic spectra of various para-aminotriphenylcarbinols in 96% H„S0,
.
They reported that each of the carbinols which was converted to its
corresponding carbenium ion upon dissolution in the acid exhibited two
intense bands in the visible region of the spectrum which had not been
found to be present in the spectrum of the parent carbinol. It is
interesting to note, however, that these workers reported no such bai
for tr '-p-d imetrhylaminophenylcarbinol in the acid and concluded in this
'.nice that carbenium ion formation had not occurred (cf. the results
6 r>
of Newman and Deno (above) with respect to their investigations on tl
ilar carbinol) . Branch and Walba also isolated a trend in band
position shifts from their data. They rationalized that increases in
re sonance stabilization of a given carbenium ion resulted in an increase
in the frequency of the band associated with that particular carbenium
ion chromophore. The simplest example of their consideration is illus-
trated by comparing the relative positions of v for the ionic speciesmax
+ + — —(C,HC)„CH and (C,H.-) C wherein v (diphenyl ion) < v (triphenyl
6 ;> 2 6 5 3 max J max r J
ion). Branch and Walba had attributed this absorption to the (C,H C )„Co J /
chromophore. Thus, these workers concluded that resonance stabilization
of the (C,H,.)„C chromophore in the trityl ion by the presence of the
"third" phenyl ring resulted in a blue shift of v . Branch and Walbamax
also contended that trityl carbenium ions exhibited two visible region
absorption bands (instead of but one) because of two carbenium ion con-
taining chromophores believed to exist in the sterically hindered parent
ion. This was in basic agreement with similar considerations made
previously by Levis e_t al_. (39) . Related studies in this area by Evans
and coworkers (42) provided evidence pointing to the existence of a
relationship between band intensity and resonance interactions in tri-
arylcarbenxum ions. These investigators reported that para substitution
of a given group on a phenyl ring in triphenylcarbenium ion resulted in
a actable increase in e for the main absorption band of that particular
don, whereas the corresponding ortho substitution of the same group
resulted in a marked decrease in e for the same band.
Studies by Dehl e_t^ al. (43), and by Deno et al. (36) served to in-
validate a number of the earlier arguments concerning the interpretation
o F the electronic spectra of trityl carbenium ions. Dehl et al.
66
oncluded from pmr investigations on various douterated triphenyl-
'.'im ions that the three phenyl rings in the cationic aggregate
were equivalent. This result suggested that analyses of the electronic
>ctra of such ions would require the attribution of any trityl car-
benium ion chromophore to the ion as a whole, therefore denying the
possibility of simultaneous existence of two (or more) chromophores in
the molecular framework of the ion. Deno e_t al. performed simple LCAO
MO calculations on aryl cations and found that the results of these
halations predicted identical v positions for the principal elec-max r r
tronic absorption exhibited by related mono-, di-, and triaryl cations.
Thus, inhibition of resonance in a sterically hindered carbenium ion
would still leave v (calculated) unchanged. Also, these calculationsmax
indicated that the intensity of the principal absorption for such ions
is invariant to phenyl ring rotation (due to steric interaction). Hence,
these authors concluded that electronic absorptions exhibited by such
ions could not be used as a measure, of steric inhibition to resonance
interaction existing within the ion. In yet a later article, Deno (44)
has again pointed to the irresolute nature of the situation in commenting
that much of the work on the electronic spectra of arylcarbenium ions
still requires revision. Olah et a_l. (45), as veil have concluded that
there is a great deal of uncertainty in the literature citations con-
cerning the visible and ultraviolet spectra of carbenium ions. Interest-
Ly enough, however, in order to rationalize some of their results
these workers employed the notion that in a sterically crowded triaryl-
rbenium ion only two of the aryl substituents are part of the absorbing
omophore, while the third functions as a cross-conjugating electron
lor or acceptor moiety. So, once again the possiblity of the existence
67
oi mult Lple chromophores in trityl-type carbenium ions is considered,
Thus, as late as 19&6, many of the problems resulting from unsatisfactory
arpretation of the electronic spectra of carbenium ions remained.
To a large degree, recent investigations in this area rely heavily
upon molecular orbital treatments of the electronic structure of aryl-
carbenium ions. Streitwieser (46:226-230) has shown that in the HMO
approximation the lowest energy transition exhibited by triphenyl-
carbenium ion can be considered to be associated with the passage of
an electron from the highest occupied bonding MO to the vacant nonbond-
ing NO in the odd-alternant hydrocarbon created upon cation generation.
To a first approximation the energy level diagram associated with this
transition is the same as for the benzyl cation, Various, more sophisti-
cated MO treatments (46:360-362) however, reflect the complex and imbro-
glio tic nature of this approach to the problem by calculating rather
grossly different charge densities for the ir-framework carbon atoms in
the ground state of the trityl ion.
By employing MO and resonance theory Waack and Doran (47) attempted
to correlate the effects of methyl group substitution in odd-alternant
anions with the resultant band shifts of the main absorption bands
found in the electronic spectra of these ions. They noted that this
type of substitution ( i.e. , methyl or alkyl) on an even-alternant hydro-
carbon had - - in the absence of steric affects - - always induced a red
shift in the electronic conjugation bands, whereas similar substitution
on a nonalternant hydrocarbon had induced either a red or blue shift
depending upon the site of substitution. They reported spectral
mges for the odd-alternant anions to be similar to those for the
ilternants and experienced qualitative success in applying the
results of their calculations to the prediction of the spectral behavior
odd-alternant ions. A particu] highlight of this effort was the
prediction thai: a-alkyl substitution on an odd-alternant cation would
result in a blue shift of v> which is in agreement with reported data.
;N rattier comprehensive MO treatment by GrinLer and Mason (48) yielded
a symmetry-based energy level diagram for structurally comparable aryl-
methyl ions. An examination of this energy level diagram revealed thot
the longest wavelength, lowest frequency transitions for a related bis-
and trisarylmethyl ion pair should be approximately identical ( cf . the
conclusions reached by Deno et_ al. (36)). However, stemming from more
acute considerations, these authors showed that the ground state
charge stabilization of a given triaryl ion was greater than that for
the corresponding bisaryl ion (recall the conclusions of Branch and
Walba (41), pp. 64-65). Thus, the highest occupied bonding MO's for
trisaryl ion were somewhat lower in energy than the related MO's
o": the bisaryl icn, and therefore, the longest wavelength transition
oi the trisaryl ion was of slightly greater energy than the correspond-
ing transition for the bisaryl ion. The dual nature of the visible
region absorption envelope of the trisaryl ions was also considered by
Grinter and Mason. Their explanation ran as follows. In the point
group "C.," (following from the anticipated propeller shape of these
ions) the highest bonding MO's of the trisaryl ions are "five-fold
degenerate, " with the attendant symmetries e, e, and a„. The MO's of
the first excited state (s) are likewise five-fold degenerate, and are
of a,, a,, a„, and e symmetries. Again, provided that the ion is
propeller shaped, transitions to the 'A™ and 'E excited terms are
allovzed and should be polarized parallel and perpendicular respective
69
co the principal Lhrce-fold symmetry axis of tlie ion. This appeared to
be in basic agreement with similar considerations which hod been made by
Lewis and Bigeleisen (49) on the phenomenon of polarization upon the
transitions in the electronic spectrum of crystal violet and malachite
green. Thus, Grinter and Mason concluded that two low energy transitions
of high intensity are expected (and are found) for such ions. (It is
here appropriate to point out that results of newer studies on the elec-
tronic absorption spectra and magnetic circular dichroism (MCD) of tri-
phenylcarbenium ions have suggested refinements of certain of the con-
siderations made by Grinter and Mason. Mo et^ al_. (50) found three MCD
bands in the near UV, visible spectrum of triphenylcarbenium ion. This
result indicated the existence of three electronic transitions in this
spectral region for trityl-type carbenium ions whereas only two such
bands were proposed to exist by Grinter and Mason. Dekkers and Kielman-
vau Luyt (51) in fact have stated that MO theory does predict three nearby
singlet * singlet transitions for a triaryl ion of D„ symmetry. Two
of these three transitions are to excited states of e symmetry and are
polarized in the x,y-plane of the ion. (This plane is defined for an
assumed coplanar arrangement of the aryl rings and the exocyclic carbon
atom. Thus, the aryl rings are perpendicular to the principal symmetry
axis of the molecular ion, the z-axis.) The third transition, however,
which is to a state of a„ symmetry and z-polarized, would not be observed
if the cation were completely planar. Since the cation is propeller-
shaped and not planar, this transition is observed (at slightly higher
•quency than the higher energy intense band), but it is considerably
weaker etian either of the highly allowed transitions to the e states.
Ls also noteworthy that these investigators were not able to resolve
70
state transitions which are observed to overlap rather tvere-
ly (maxima arc at 23.2 and 24.6 kK respectively with a reported c of
38,700 for each band (45)). This suggests an inherent relationship
Lween the electronic states in the ion from which those two bands
originate. Consequences of this implication concerning the chromo-
phoric nature of triairylcarbeniuni ions lie in the forthcoming text,
vide infra.
It is now interesting and profitable to examine the wave functions
for the molecular orbitals which were considered by Grinter and Mason
to correspond to the energy levels which arise upon generation of a tris-
arylmethylcarbenium ion. The forms of these wave functions are:
i>>. = a<> + bOJi , + i> ., + ij> lH) {5}1 'c raryl raryl aryl
r II aryl raryl
and
$__, - (iji . +i>
. ,-
2<J> 1lf)//6 {7}II aryl aryl varyl
where (j> is the wave function of the 2p state of the exocyclic carbon
atom. The ij^ T functions represent the highest occupied bonding states
of the ion, and their forms indicate that the bonding contribution of
the "third" aryl ring (iji ,,) is of principal significance in tyTT i«
ite: The basic forms of these wave functions are virtually identical
to the corresponding wave functions used in the calculations by Dekkers
Ilan-van Luyt (51).) Since the degeneracy of the ty__ functions
71
, >en removed to an appreciable degree by virtue of the nonplanarity
of the ion (48) and by configuration interaction effects (46:227), (51),
it follows that only one of the long wavelength, low energy transitions
should reflect significant electronic contributions to carbenium ion
stability (or instability, as the case may be) by the so-called "third"
ring which is denoted above as aryl" in equations {5} and {7}. This
consideration is given additional substance from the results of various
investigations. Barker and coworkers (52) studied the electronic spectra
of derivatives of malachite green produced by substitution in the "non-
anilino" phenyl ring. They showed that the longest wavelength absorption
band recorded for each derivative reflected primarily a flow of electrons
from the two para-N , N-d imethy1 substituted rings towards the exocyclic
carbon atom. This is in keeping with MO calculations which have estab-
lished that this carbon atom bears the principal degree of positive
charge in the ground state of the ion (53) , (an expected result) . There-
fore, replacement of the para-N , N-dimethyl groups with poorer electron
releasing substituents resulted in a blue shift of v . These workersmax
also demonstrated the existence of a linear relationship between the
appropriate Kammett constant for the phenyl ring substituent and the
magnitude of shift in v induced by that particular substituent. Thus,max J *
a type of cross-conjugation seems to exist between the phenyl ring and
the remainder of the conjugated system with respect to the energy of
I longest wavelength transition. The results of studies by Hopkinson
and Wyatt (54) concerning substituent effects upon the electronic ab-
rptionsof phenolphthnlein monopositive ions (Figure 5) allowed these
workers to conclude that v of the second longest electronic transitionmax &
exhibited by these ions reflected primarily a shift in electron density
72
.
',. "third" ring towards the exocyclic carbon atom. In phenol-
halein this "third" ring is the ring to which the ortho-carboxylic
acid group is attached. Hopkinson and Wyatt also compared the elec-
ronic absorption spectra of phenolphthalein and phenolsulphonphthalein
and observed a considerable blue shift of the second baud for the
"sulpho" containing phthalein. This was expected owing to the apprecia-
ble electron withdrawing power of the sulphonic acid group. These results
were corroborated via extended HMO calculations to resolve the electronic
effects which arise from para-substitution of u-electron donating groups
oq two of the phenyl rings in triphenylcarbenium ion. Furthermore, the
results of these calculations were found to be in agreement with the
results of the HMO calculations which had been carried out by Mason and
Grinter (48). These calculations also verified that such substitution
of "-electron donating groups served to remove still further the de-
generacy of the two highest occupied MO's in triaryl-type carbenium
ions (see. p. 70) with the higher energy occupied MO acquiring a greater
iW
R = -C00H
R' = -CH,, -CI, -Br, etc,
R" - -CH-, -CI, -Br, etc,
Fig. 5. Phenolphthalein monopositive ions
73
Lectron contribution from the rings which bear the more potent elec-
tron releasing substituents. Consequently, the conclusions reached
which qualitatively associate the two principal electronic absorptions
of triarylcarbenium ions with the MO's from which these transitions
originate (p. 70) have been upheld. In addition, Hopkinson and Wyatt
also isolated from their spectral data a linear variation between
Hammett c-meta substituent constants and shift magnitudes of the higher
energy electronic band produced upon the. substitution of groups ortho
to the ring hydroxyl groups in phenolphthalein. Thus, whereas the
results obtained by Barker and coworkers (52) indicated a cross-con-
jugation effect to exist upon the frequency of the lower energy band
and the "third" aryl ring, Hopkinson and Wyatt showed a similar effect
upon the frequency of the higher energy band by the two rings in a
given triaryl-type carbenium ion which are the primary (as compared
to the remaining ring) ir-electron donating moieties.
Finally, it is still not clear whether the high intensity electronic
absorption bands exhibited by triarylcarbenium ions may or may not be
considered rigorously as charge transfer transitions! This concern is
not crucial to the considerations made in this work; but for complete-
ness a few remarks shall be tendered. Initially, in this discussion of
electronic spectra, it was assumed rather tacitly that these electronic
bands are charge transfer in nature (pp. 61-62). However, these ab-
I
tions do not meet with certain of the criteria (55) which have been
employed for classifying electronic transitions as "charge transfer."
1 h (56:65-66) for instance, has pointed out that the factors tending
to indicate charge transfer interactions in monoaryl tropylium ions
74
(prov« d by Couch to exhibit intramolecular charge transfer) are either
not present, or are opposite, in mononrylcarbenium ion5;. Furthermore,
related studies by Couch (57:113-122) have suggested that triarylcar-
benium ions as well do not exhibit bona fide charge transfer interactions.
Dauben and Wilson (58) have at last demonstrated the existence of au-
thentic charge transfer interactions for systems containing triaryl-
carbenium ions. They accomplished this via the preparation oi" various
pyrene-triarylcarbenium ion complexes wherein the coordinated carbenium
ion was found to function as a particularly potent ii-aeceptor. The
ctronie spectrum of any of these complexes gave a very intense band
at. relatively low energy (viz. 14.1 kK for coordinated triphenylcarbenium
icn) which was not present in the spectrum of either component. These
results imply that the existence of a "true" charge transfer inter-
action in a system requires an appreciable shift of electron density
from a specific location in the ground state of the parent molecule
(ion, etc.) to a new (and removed) location in the excited state.
Perhaps then, it is in fact not extremely unrealistic to treat the
principal electronic absorptions of arylcarbenium ions as charge trans-
fer. Ramsey (55) has alluded to this assumption by suggesting that a
charge transfer cransition in a triarylborane can be associated with
the promotion of an aryl ring ir-electron into the empty available p_
orbital on the boron atom. Similarly, since the results of various
studies on the electronic spectra of arylcarbenium ions have associated
the main absorption bands with transfer of aryl ring 77-electron
density to the exocyclic carbon, those transitions may, at least broadly,
I ;ed as charge transfer. Arguments contrary to this class L-
Lon can be registered based upon degree of charge transfer. For
75
19tance, work by Olah et al. (59) on the pnir and "" F nmr spectra of
I fluorocarbenium ions has demonstrated that the degree of charge
derealization into the aryl rings in these ions is substantial. Thus,
alectronic ground state charge derealization in these ions appears
tc be considerable. Extended HMO calculations (54), however, have
indicated an appreciable difference in carbon atom charge densities
between the ground and first excited states in arylcarbenium ions.
Let it suffice, therefore, to say that, this aspect of arylcarbenium ion
electronic spectral investigations remains largely a moot issue.
An examination of the electronic spectra and attendant data collected
during this work is now in order. Examples of electronic spectra ob-
tained for carbenium ions derived from a related series of compounds,
name!} pyLOH, Pd (II) (pyLOH) (LN)C1 2> and Pd(II) (pyL011)
2Gl
2(where pyLOH
equals 4-pyridyl-4-methylphenyl-4-fluorophenylmethanol, and L equals
diphenyl-4-pyridylmethane) , have been presented in Figures 6 — 8, p. 76.
As these spectra are representative of all electronic spectra recorded
for the carbenium ion species considered herein, no other electronic
spectra are presented. Pertinent electronic spectral data are given
in Table II, pp. 77-78).
The carbenium ion electronic spectra obtained in this work are
found to be very similar to the related spectra (in the same spectral region)
which have been reported by previous investigators (Richardson (6) and
Wentz (8)). An examination of these spectra reveals the presence of two
broad, intense absorption bands located within the 33.3 — 18.2 kK range.
The lower energy band found in the spectrum of a given carbenium ion is
ys the more intense. The e values reported in Table II indicate
76
A .
5
o.o 420.0 18.2 17.4
1.0 _,
25.0 22.2v (kK)
Visible spectrum of the carbenium ion derived from 4-pyridyl-4-mef.hylphenyl-4-fluorophenylmethanol, in 70% HCIOa. v ,20.2,29.7.
maX
A 0.5 „
o.o X33.
Fig. 7.
*-
i i
28.6 25.0 22.2 20.0 18.2 17.4V (kK)
Visible spectrum of the carbenium ion derived from Pd (II) (pyLOH)
(Inr)Cl2j where pyLOH is 4-pyridyl-4-methylphcnyl-4-f luorophenyl-methanol, in 70% HCIO4. v"max , 20.6, 27.6.
A 0.5 )
20.0 18.2 17.4T"
25.0 22.2
V (kK)
Fig. 8. Visible spectrum of the carbenium ion derived from Pd (II) (pyLOH^Cloj where pyLOH is 4-pyridyl-4-methylphenyl-4-f luorophenyl-
thanol, in 70% HCIOa. "S , 20.6, 27.7.^ max
.
.
LOc--i
4-J
cd
stOi-H
y
o--•
•H
mCD
Hoa)
P.
poH
Ha)M X>M Ptd
CD CO1—
-!
,X3 0)
cd PH O•HpTO
>
.c4-J
V!
oU-l
ctf
OHsd
Puocu
p.</)
U-HPOP4-J
UCJ
rHw
P
•dci
T3P301
uXH
X)m
o Ot-i .Ho uc_> WC >-.
o o
CJ -H
s
79
. directly. This band is the so-called ":•:" electronic band so
classified by Lewis and Calvin (60) in an elegant work on the color of
garlic compounds. Similarly, the higher energy, less intense (in this
cast-) band, is labeled the "y" band (60). Certain relevant trends are
established by the spectral position shifts of these main absorption
bands which occur upon proceeding from R = phenyl to R = 4-raethoxyphenyl
(see Table II) for a related series of carbenium ions; and from the
comparison of the spectrum of a free alcohol carbenium icn to that of
the corresponding complexed carbenium ion. An inspection of the
resnective v values (Table II) reveals these trends to be themax
following:
(i) For carbenium ions derived from a family of precursors (e.g.,
the 2-pyiidyl alcohols, the 4-pyridyl alcohols, etc.), in proceeding
from R = phenyl to R = 4-methoxyphenyl v of the x-band decreases ca.* Jr J max
0.4 — 0.5 kK for the free alcohol ions and ca. 0.9 — 1.0 kK for the
complexed alcohol ions. And for the 4-pyridyl ions the x-band v forr rj max
a given free alcohol ion is lower than that for the corresponding
complexed ion, with the difference in related x-band frequencies dimin-
ishing in going from R = phenyl to R = 4-methoxyphenyl. Thus, whereas
the x-band v is at 20.7 kK (free alcohol ion) and 21.3 kK (bis-max
complexed ion) respectively for R = phenyl, it is at 20.2 kK (free
alcohol ion) and 20.3 kK (complexed ion) for R = 4-methoxyphenyl.
(ii) Making the same comparisons as in (i) (above) on the relative
y—band v values reveals that proceeding from R = phenyl to R = 4-
hoxyphenyl results in a concomitant increase in v of ca. 0.7 kK:J
r
J max —but this does not include the free 4-pyridyl ions where an increase in
y-band v cf only ca. 0.3 kK is encountered. And, in this case themax —
80
for a given 4-pyridyl ion is found to be higher (usually
1.0 l-.K) than the y-band v of the corresponding completed ion.max * ° '
however, no apparent trend exists in the magnitudes of observed
difference in y- band vrvalues for a related free ion — ci :ed ion
TT13.X J
pa ir.
(iii) A comparison of the x- and y-band v values of a fro*-max
2-pyridyl ion and corresponding 4-pyridyl ion shows that for a given R
group vris always lower for both the x and y absorptions. (Note:
This particular trend was also exhibited by a related series of 2-thi-
azolyl \^s. 5-thiazolyl earbenium ions (8). That is, the x and y ab-
sorotions exhibited by a 2-thiazolyl ion were always found to be at
lower v than the same absorptions for the corresponding 5-thiazolylmax
ion.
)
For convenience and simplicity these spectral trends are summarized
in the immediately succeeding statements. The effecu of proceeding from
R - phenyl to R = 4-methoxyphenyl is reflected by a bathochroraic (red)
shift of the x-band and an hypsochromic (blue) shift of the y-band.
And, the effect of coordinating the earbenium ion always results in an
hypsochromic x-band shift and a bathochromic y-band shift. It can
now be shown that these results are qualitatively in accord with con-
siderations made previously concerning the electronic spectra of aryl-
carbenium ions. For instance, if the energy of the x-band transition
exhibited by pyridylcarbenium ions is dependent primarily upon a flo
of electrons towards the exocyclic carbon atom from the aryl group(s)
h are of predominant electron releasing capability, it is expected,
and found, that the energy of this transition is reduced upon replacing
R = phenyl with R - 4-me'Lhylphenyl, or R = 4-methoxyphenyl. The
81
subsequent cross-conjugationa] effect of this R substitution on the
energy of the y-band transition is illustrated by the corresponding
band blue shift in the spectra of a related series of ions. More
important to this work, however, are the band shifts which take place
as a consequence of complexation of a given carbenium ion. It follows
that if the transition energy of either or both of the bands (x, and
or, y) in the spectrum of a pyridylcarbenium ion is associated, at
least to a degree, with a flow of electrons from the pyridine ring to
the exocyclic carbon atom, any change in the electronic nature of the
pyridine ring making it a more effective electron releasing moiety
should result in a lowering in energy of that (those) electronic band(s).
Furthermore, it: is reasonable to expect that the coordinated pyridine
ring would in fact be a better donor than the pyridine ring in the
uncomplexed ion as in 70% HC10, the ring nitrogen is certainly protonated
in the free ion, vide infra. Clearly then, any difference in these
situations should reside in the fact that in the. coordinated pyridine
c?se. electrons flow from an ostensibly neutral ring towards the exocyclic
carbon; whereas in the uncoordinated pyridine carbenium ion electrons
are required to flow from a ring which already bears a positive charge
(the proton) resulting in a comparably unfavorable energetic trans-
formation. Now, since coordination of a given pyridylcarbenium ion
results In a blue shift of the x-band, and a relatively substantial red
shift of the y-band, it appears to be the case that the y-band transi-
tion energy is, to an appreciable degree, paronymously related to the
pyridine ring, and therefore dependent upon its electronic nature.
These arguments, of course, serve to indicate that the pyridine ring
is the so-called aryl" which appeared in the wave equation, i> = {5},
I i|)__, - [7K presented previously (p. 70). So, from a qualitative
tint it is reasonable that the ir-electron energy level of;J'-fT ,
would be lowered with the pyridine ring protonated relative to the energy
level of ^ T -nwith this ring coordinated, as in the protonated situation
pyridine ring would be expectcdly more electronegative. Hence, if
energy level of \\> is not altered as much as that of 'J'-r-rt for each
of tuese. two possibilities, the y-band should (and does) shift red for
the coordinated carbenium ion relative to the "free" ion. It is also
suggested that these assessments are correct owing to the magnitude of
shift in energy observed for the y-band upon pyridylcarbenium ion co-
ordination. An inspection of the v data (Table II, pp. 77-78) for the
4-pyrldyl ions reveals that coordination induces a red shift in the
energy of the y-band proportional to as much as 2.1 kK for the palladium
plexes of the R = 4-methylphenyl ion y_£. the corresponding "free" ion.
Similarly, for this ion, the x-band is blue shifted only 0.4 kK. (This
trend is also realized by the remainder of the data found in Table II.
Furthermore, related data reported previously by Richardson (6) and
by Wentz (8) support this trend.) Consequently, the electronic energy
changes which are produced in the ion via coordination appear to be
related principally to the attendant changes in the transition energy
of the y-band. This engenders the speculation that relative energetic
contributions provided by coordinated metal species towards stabilizing
"complexable" carbenium ions would be reflected in a comparison of the
respective y-band transition energies for a given complexed ion. Per-
re work will furnish this possibility with substance.
ility in HCIO^ — H?Q as Determined by the "Peno'[
ion Method . The suitability of the Deno titration method (10,11)
for the determination of the thermodynamic stabilities of the carbenium
qs investigated in this study has been aptly demonstrated by Wentz (8),
The. interested reader is referred to this work for a salient discussion
of the necessary and pertinent experimental considerations. Reagent
perchloric acid> 70% HC10, , has proved to be a most appropriate ioniza-
tion medium for these titrimetric stability determinations. It is
a potent mineral acid as reflected by its thermodynamic ionization
constant (K ) reported to be. 3800 (61). In fact, as pointed out by
Gillespie (62) , HC10, — H,,0 systems can be more acidic than reagent
H_S0., and only certain, nonaqueous superacid systems afford higher
acidity than aqueous HC10, . Indeed, HC10, — HO is found to be uniquely
between neat H„S0^, and superacid systems such as HSO^F — SbF (used
by Olah (63) to prepare many relatively unstable carbenium ions) as a
useful solvent for the generation of trityl-type carbenium ions.
Freedman (64:1535) has commented that carbenium ion investigations in
concentrated IUSO^ can be complicated by side reactions such as sulfona-
tion or oxidation which in turn can destroy either the parent carbinol
or the carbenium ion. Problems such as these are without a doubt respon-
sible for much on the confusion found present in earlier studies on
carbenium ions. The superacid system HSO^F — SbF,. as well, recently has
been shown to generate carbenium ions as a consequence of oxidation by
SbF or S0„ (65). Thus, previously devised methods and mechanisms of
carbenium ion generation in this solvent may require extensive modifica-
tions. Furthermore, the use of superacids for the preparation of these
•s of carbenium ions would certainly require modifications of the
R4
mo lit':ni ion method to take into account the fact that superacid media
ar« nonaqueous. Therefore, at the minimum, the definition of new acidity
nction parameters would be requisite, as well as drastic alterations
he mechanics of the aqueous titration technique.
A consideration of the equilibria which pertain upon carbenium
ion generation is in order. An examination of the literature in this
area reveals that on occasion various authors are wont to write H as
the acid species responsible for the com yion of carbinol to carbenium
ion. Albeit convenient, this practice is certainly not rigorous, and
can frequently be misleading. In the present situation where perchloric
acid has been employed as the ionizing medium it is necessary to choose
between HC10, or H-0 as the principal proton source towards the ion
precursor carbinols, assuming, of course, that other complex or unusual
acid species do not exist in this system in appreciable concentration.
Since concentrated solutions of such mineral acids as H~S0, , HN0~, orIk 3
HC3 , are not capable of carbenium ion generation in instances where
HC10. - HO is, it follows that HC10. is the acid species responsible
for carbenium ion formation in those systems in which these other
acids produce little or no carbenium ion. As a result of nmr studies
on KC104
- H20, Redlich and Hood (66) reported that reagent 70-72%
IIC10, (ca. 11.8 M) is approximately 75% ionized. Hence, sufficient
molecular UC30, is present in 70% HC10, — H„0 for carbenium ion genera-
tion in systems which contain as solute limited quantities of ion
+arson pyridylmethanol. And here, concentrated solutions of 1U0
do not convert carbinol to carbenium ion to any appreciable extent,
exi ! for the relatively more stable cations such as those which contain
ly electron releasing phenyl ring substituents such as 4-methoxy
85
ups. Thus, the ionization of triphenylcarbinol in HC10, — H?
may
1<^ written appropriately as:
(C6H5) 3
COH * 2 I1C104
* (C6H5)C+
+ H.^o"1" + 2 CloT {8}
And, even though ring nitrogen protonation of the free alcohol ions
tends to complicate the attendant equilibria, the pyridylmethanols
should be ionised similarly. (Also, see the discussion which ensues
(p. 98) in reference to the stability data contained in Table V.)
The equation for the Deno acidity function (H ) may be written as
\ ' P'V + ^"W »)
i here the values of IL for a particular acid medium are a measure of
the capability of that medium at various concentrations to ionize a
given alcohol (R-OH) generating the corresponding carbenium ion (R ).
Since this equation is of the form, y = b + mx, it follows that if a
series of values for H are known, and if the respective concentrationsK.
of R-OH and R for a related alcohol — carbenium ion pair can be ex-
perimentally determined at the different H^'s, pRp-f may ^ e obtained
as a quantitative measure of the thermodynamic stability of a given
carbenium ion. Of course, it. is required therefore that the acid medium
be capable of measurably ionizing the alcohol over a range of acid
centrations; and it is also inherently required that in order for
the H relationship to hold, the slope (m) of the curve got by plotting
HD vs. the log term be equal to 1.00. Wentz (8) showed that both of
86
Lons were met in HC10, - HO for the pyridylmethanols
'. ar< considered in this work.
A list of R values for aqueous HCIO^ and the corresponding wt %
Ld are given in 'Cable ITT, p. 87. A plot of -H vs. wt % acid
(Figure 9, p. 88) yielded a smooth curve of approximately constant
slope which was easily extrapolated (as shown in Figure 9) to 70.0%
HC10, ir: order to obtain 1L values for acid concentrations greater than
60.0%. Employment of the fact that Beer's law (A = ecb) directly
relates the absorbance (A) of an absorbing species to its molar con-
centration (c) , allows values for [R-0H]/[R ] in equation {9} to be
obtained by measuring the absorbance (to ±0.001 absorbance units) of the
carbon iuin ion at various HC10, concentrations. The absorbance (con-
centration) of R-OH was then taken as the difference between the Beer's
law absorbance of the carbenium ion (see Dilution Curves, p. 93) and
the actual absorbance of the ion at a given acid concentration. This
method of data treatment is valid since R-0H does not absorb at Xmax
of the carbenium ion. Consequently, it is justifiable to assume that
i the actual absorbance of the carbenium ion matched the expected
absorbance as predicted from Beer's law, the alcohol was ionized to an
extent of ca. 100%. Naturally then, as the ionizing acid was systemat-
ically diluted (actually, at the onset of the titration molecular HC10.
is also converted to II_0 , CIO.) by the addition of measured increments
of water, the absorbance fall off was linear (Beer's law dependent) as
as the alcohol remained essentially 100% ionized. However, as soon
as the carbenium ion became titrimetrically reconverted to alcohol pre-
cursor as a result of further addition of water, absorbance fall
no longer linear; and indeed, it was greater than that predicted
37
Table IIIValues of H in Aqueous HC10. at 25'
-\ Wt % HC104
a
3.79 30.0
4.61 35.0
5.54 40.0
5.95 42.0
6.38 44.0
6.82 46.0
7.31 48.0
7.86 50.0
8.45 52.0
9 . 05 54 .
9.68 56.0
10.37 58.0
11.14 60.0
aDeno et al. (11)
r-1
3^
a
89
from octrapolated Beer's law straight line. An examinatiun of the
dilution curves (Figures 10-12, p. 93) illustrates this straightforwardly,
Thus, for each carbenium ion so titrated, a collection of absorbance
data was obtained as a function of changing wt % HC10,
.
Tata treatment was carried out using the methods employed by Wentz
(8) with some minor modifications. These methods may be described con-
veniently in conjunction with a stepwise examination of the pertinent
arithmetic relationships required for data treatment. Initially, an
unweighed .sample of ion precursor pyridylmethanol (free or complexed)
is dissolved in sufficient reagent HC10, (determined as 70.87%, see
below) co produce an acceptable on-scale (viz., 15-25% T) spectrophotom-
eter reading at X for the absorbing species (the carbenium ion). Themax
total solution sample is now weighed and the absorbance recorded. Then,
measured increments of deionized water are added to the carbenium ion
solution and, after thorough mixing, the absorbance of the sample is
reread. The wt % of the acid solvent resulting from dilution is calcu-
lated from:
Wt % acid = - t ^ f
WVg)+°
f
T)\ * H n -AA-A {10 >wt (g) of sample + st (g) of H„0 added
where wt of acid == original wt % acid (70.87%) x original total sample
wt. Thus, the wt % of the. acid can be determined following each dilution
by simply noting the cumulative quantity of water which has been added
to chat stage in the titration. The volume of the sample following
h addition of water is then determined from:
90
pie volume (ml)cumulative sample wt (g)
corresponding p (g/ral) of the sample
>ecific gravity of the acid solvent. And therefore, it is
obviously necessary to have values of p for aqueous RCXO,. (Wentz had
obtain.,., l this information from a plot of p (4 sig. fig.) vs. wt % (4
. fig.) for solutions of aqueous HC10, (see Table IV, p. 91).)
However, since the volume of an aqueous HC10, solution does not increase
linearly upon dilution with water ( i.e. , the volumes of water and parent
acid solution are not additive), Wentz: found it necessary to "blow up"
this plot in the region of each wt % datum point in order to obtain
corresponding value for p. This procedure was found to be extremely
tedious owing to the difficulty found in obtaining 4 sig. fig. accuracy
(to insure reliability of related data to 3 sig. fig.) for a considerable
amount of wt X data from such a graphical readout. This particular
problem was conveniently alleviated by taking a "least squares" curve
fit of Brickwedde's data (67), (Table TV, p. 91) to yield a "printed
out" series of values for p spanning the range 0.00 wt % HC10, to 75.004
wt % HCIO,. (The details of this least squares treatment are given in
the Appendix.) Thus, having experimentally determined p of the original
reagent acid, the corresponding wt % of the original acid was obtained
from this p — wt % tabulation. The use of equation {10} then yielded
wt % da La for subsequent dilutions, in turn for which corresponding p
values were available from the p — wt % tabulation. Finally, it is
: iry to determine the dilution fraction (D.F.) for each dilution.
This is obtained from:
9.1
Table IV
Specific Gravity of Aqueous HCIO, Solutions at 25'
Specific Gravity (g/'ml)
a
0.997
1.026
1.056
1.088
1.123
1.160
1.200
1.244
1.291
1.343
1.400
1.462
1.527
1.596
1.664
Brickwedde (67)
wt %
n „ origina] sample volume (ml)
cumulative sample volume (ml) '
Ln, inspection of equation {9} reveals that the required values for
the log term are ratios of concentrations (of alcohol to corresponding
carbenium ion), and not absolute concentration values. Thus, it is
not necessary to determine the concentration of the absorbing species
at any time during the titration. It is only necessary to measure
absorbances throughout the titration at the various, known dilution
fractions, which are proportional to changes in the concentration of the
absorbing species. Hence, a plot of absorbance (A) vs . dilution fraction
(D.F.) yieJ.dc a plot which reflects the decrease in concentration of
carbenium ion as a consequence of dilution as well as of reconversion
to alcohol precursor. And, the only requirement which must be met in
ord >r for this treatment to hold is that the alcohol be ionized com-
pletely at the onset of the titration. An examination of the dilution
ves for free and complexed carbenium ions derived from the same al-
cohol precursors (Figures 10-12, p. 93) clarifies these considerations.
In each cr.se the beginning of the titration ( i.e. , in the vicinity of
D.F. = 1.00) corresponds to a linear fall off of absorbance with dilu-
i. Thus, at this juncture of the titration the alcohol precursor
^p.'cies is essentially completely ionized; and, to repeat, the effect
of the addition of titrant (water) is to titrate molecular HC10. while4
only diluting the carbenium ion. When the equilibrium reconversion of
ion to alcohol (described by equation {1], p. 9) becon-s mea-
rable the titration curve begins to slope down and away from the
lated Beer's law straight line as seen in the dilution curves.
,;"CJ
0.40
0.80 0.725D.F.
Pig. 10. Dilution curve for the titration of 4-pyridyl-4-methylphenyl-4-fluorophenylcarbenium ion
0.6
(i.40
0.725D.F.
Fig. 11.
0.60
0.40-
,+Dilution curve for the titration of [Pd (II) (4-pyL) (Ln)C12j ,
where 4-pyL = 4-pyridyl-4-methylphenyl-4-fluorophenylcarbeniumion
0.90 0.80 0.725D.F.
Fig. 12.2+
Dilution curve for the titration of [Pd(1 1) (4-pyL) 2CI2]'where 4-pyL = 4-pyridyl -4-methylphenyl-4-f luorophenylcarbeniumion
' isr straight line shown (which practically bisects the plot
horizon corresponds to the so-called "1/2" Beer's law line. This
cond line ha i b sen urawn by considering that if at the onset of the
•i :ration the concentration of the absorbing species (the carbenium
ion only) were divided by two, its absorbance would also be. divided
by two. Thus, this "1/2" Beer's law line may be constructed from a
series of points got by halving the absorbance values which served to
establish the initial (upper) Beer's law line. Consequently, the
intersection of this second Beer's law line with the dilution curve for
a given carbeTiium ion marks the D.F. location at which [R-OH] is ex-
pected to equal [R ] during the titration of that ionic species. The
point of this graphical treatment resides in the fact that the log term
in equation (9} is equal to zero for equal values of [R-OH] and [R ]
.
Hence, vertical extrapolation from this intersection point to the ab-
scissa of the plot yields the D.F. at which the corresponding value of
EL equals pKi of that carbenium ion. Of course, H is obtained after
relating D.F. to the appropriate wt % HC10,, (note: here, simple
arithmetic interpolation is usually necessary to obtain the required
values for wt % HClO, since the data points used to plot the dilution
curve seldom include the exact D.F. values at which [R-OH] = [R ]),
which in turn is related to H (see Figure 9, p. 88).K
The thermodynamic stability data resulting from these titrimetric
mts upon the pyridylcarbenium ions investigated in this study
are pr I in Table V, pp. 95-96. The following considerations which
are in respect to this data are relevant.
The reliability of the data was corroborated through a titrimetric
cnient of pKR+ for triphenylcarbenium ion. The value obtained
a
96
V10)-)
d0J
4J
Xw
.-I
cd
H
g
<l
rHid
o
o00
udoc>>-<
ft
do
cOJX>-i
td
ONVO
vcvO
ONCM
Ovo
vo
<3-
~cj-
cr> vO
Or-«
COin
cmrH
y
oJnI
-J-
13ft
OJ
QJ
:
rH Uft ^-.
O (J
13
coj
XftI
II
Pi
>,a<u
XftH>.X4_)
CD
EI
~cr
I
CMCO
ft
97
ile V) is in good agreement with pK . measured for this ion in 60Z
KC10. by Bono and coworkers (11) using the titration technique.
;\ requirement which obviously must be met by the complexed carbenium
Ions to insure that the corresponding stability data be meaningful is
that the coordinate bond between the palladium metal center and the
pyridine ring remains intact throughout titration. Undoubtedly, the
protonic solvent will compete strongly for the basic ring nitrogen site(s)
as evidenced by the fact that the pyridine ring is protonated for the
"^ree" alcohol ions in 70% HC10, (shown below). The simplest demonstra-
tion of the stability of the coordinate bond during titrimetric (and
1 QJ""'F nmr chemical shift) measurements stems from the fact, that the
electronic spectrum of a given complexed 4-pyridylcarbenium ion in HC10,
— HO remains unchanged for a period of at least a few hours. In fact,
the electronic spectrum of the bis complex of 4-pyridyl-4-methylphenyl-
4-fluorophenylcarbenium ion in HC10, — H„0 was found to remain intact4 2
for 12 hours as evidenced by the fact that during this tine no appreciable
change in band intensity could be observed; and, more importantly ab-
solutely no shifts of the x- and y-absorption bands had taken place.
Furthermore, the solution color of the complexed ion did not deteriorate
to the solution color of the corresponding "free" ion. Additional
investigations on the intact nature of complexed carbenium ion electronic
i rum in HC.10, — il o upon the lone 2-pyridylmethanol complex isolated
(see Experimental, p. 36) revealed that as soon as this compound was
dissolved in the acid the intensities of the x- and y-absorptions began
to increase steadily; and also, the x- and y-absorptions began to shift
diately towards the respective positions of these absorptions
characteristic of the uncoordinated, "free" 2-pyridylcarbenium ion.
98
lylcarbenium ion i
' ,: waa insufficiently stable to
be studied using HCIO^ -• HO as an ioniz sdiura- A comparison of the
stability data for a related free ion— complexed ion series also
:ves to establish that the metal — nitrogen bond(s) remain(s) intact
during thermodynamic measurements, and that coordination certainly
bilizes a given carbenium ion relative to the corresponding free ion.
An examination of the respective AG° values (Table V) indicates that
the degree of stabilization afforded by coordination is greatest for the
4-pyridylphenylcarbenium ion. Although the thermodynamic titration
data for the unsubstituted phenyl carbenium ions is somewhat suspect
owing to rearrangement of these ions in HC10, — H^O (discussed, pp. 103-
19106) j F nrar chemical shift measurements in conjunction with relevant
fr« e energy relationships substantiate the reliability of the reported
stability data, vide infra . The stabilizing influence of coordination
upon a given carbenium ion is also seen on inspection of the dilution
curves fur a free ion — completed ion related series (see Figures 10-12,
p. 93). Here, the Beer's law dependence of the plot reflects the
relative stability of a given carbenium ion. That is, a relative com-
parison of the D.F. values corresponding to that point in the titration
at which each dilution curve no longer adheres to Beer's law, allows an
ordering of the stability of all carbenium ions investigated by the
titration method. This particular D.F. for the free 4-pyridyl-4-r>ethyl-
phenyl-4-fluorophenylcarbenium ion is ca. 0.9?., whereos the corresponding
s for the complexes of this ion are ca . 0.87, thereby showing
i it the coordinated ions are more stable.
a stability constant values (K) found in Table V correspond to
iquilibria which are est l1' d upon the titrimetric reconversion
99
of a particular pyridylcarbeniura ion to its free or complexed precursor.
equations for these equilibria have been presented previously (p. 16)
but for expedient purposes they are repeated in the order in which they
are discussed. The equilibrium titration of a free alcohol ion is
represented by:
H+pyL
++ 2 H
2t H
+pyLOH + H,0
+{2}
This statement is consistent with the fact that the unionized alcohol is
protonated. This is certainly expected for pyridine in such strongly
acidic media as 45-70% HC10, (these values correspond to the acid con-
centration ranges spanned during the titration of the more stable car-
benium ions). Moreover, Wentz (8) demonstrated that equation {2} was
the correct equilibrium statement for describing the titrimetric re-
conversion of an uncoordinated thiazolylcarbenium ion to its alcoholic
precursor. This was accomplished by titrating the N-methyl iodide salt
of a particular 2 > 4-dimethyl-5-thiazolylcarbenium ion yielding a K value
which was identical to that which he had obtained for the titration of
the original alcohol ion. Clearly then, in each instance the thiazole
ring nitrogen must have borne a unit positive charge throughout titra-
tion. Also, by assuming that the pyridylmethanol pyridine ring is of
-9similar basicity as pyridine itself (K, pyridine = 2.3 x 10 ) , it
n be shown from simple acid-base equilibria that a given pyridyl-
methanol dissolved in 1 H H.,0 at an initial alcohol concentration of
03 "., is protonated to an extent in excess of 99.99%! (This calcula-
i ion is given in the Appendix.) Since, for the HC10, — H„0 — pyridyl-
loI systems considered, the combined concentration of IIC10. — H_0k 3
is al gn ater than 1 M, and the concentration of a pyridyl-
small comparatively, the pyridine nitrogen is protonated
. 100%) for a free alcohol carbenium ion. A direct effect of ring
;en protonation on carbenium ion stability is indicated from a
parison of the data in Table V for a structural!}' related pair of
e '.'. and 4-pyridylcarbenium ions. That is, a 2-pyiidyl ion is seen
to be considerably less stable than its 4-pyrLdyl congener. The only
significant difference, between such a related pair of cations should
reside in the disproportionate degree of positive charge separation in
the 2-pyridyl vs. the 4-pyridyl ion. Hence, it appears that the de-
stabilizing consequence of like-charge repulsion on the stability of a
given 2-pyridyicarbenium ion is relatively substantial in that the
positive charge en the 2-pyridyl nitrogen is in much closer proximity
to the carbenium ion center than in the corresponding 4-pyridyl ion
situation.
The equilibrium titrations of the coordinated carbenium ions may
be represented by:
Cl„(L>7)rd(TI)pyL
+ +2H.0 ± Cl o (LKT)Pd(II)pyL0H + H„0
+{3}
and
Cl2Pd(Il)(pyL)2
++ 4 H
2t Cl
2Pd(II)(pyLOH)
2+ 2 1I
3
+{4}
Statement {3} designates the reconversion of a singly charged coordinated
ion to its complexed "mono" alcohol precursor. The stability
ited with this t ran:- format! on therefore afford a reflection
I
>' the stabilj ' effect exerted by the metal on a single pyridylcar-
l.un Lon. Statement {4} corresponds to the reconversion of a "bis"
benium ion complex to its alcoholic precursor and is premised npon
complete ionization of both of the coordinated alcohols in 70% HClO,h
( .e. , prior to titration). The £ data (Table II, pp. 77-73) indicate
that both coordinated alcohols are ionized for the bis complexes in the
19acid. The F nmr chemical shift data (vide infra ) also corroborate
this consideration stemming from the fact that the absorption position
19oi the F nmr signal for a mono complexed ion is identical to that
found for the corresponding bis complexed ion. If both the coordinated
alcohols in the bis complex Xvrere not ionized, either a time averaged
signal would be expected or two signals would be observed in the spectrum
corresponding respectively to an ionized and an unionized ligand raole-
19cu.'e. That is, the F nmr spectrum should reflect the presence of
either two rapidly equilibrating, or two distinctly different, fluorine
nuclei. Furthermore, as previously indicated, the dilution curves re-
sulting from the titrations of the bis complexed ions all exhibit
appreciable Beer's law dependency at the onset of the titration. If
the bis complexes had not been ionized completely upon dissolution in
70% HC10, , an immediate lv detectable equilibrium would have been estab-
•had upon addition of the initial quantities of tit rant (water) , and
no Beer's lav.7 dependence would have been illustrated by the Beer's law
Lot at the beginning of the titration. This consideration alone does
no 1 preclude the unique possibility of having ionized only one of th
irdinated alcohols in a bis complex in 70% 11C10, . However, the e data,
which indicate multiple ionization of the bis complexes in the acid,
vide rathor forcing evidence substantiating the assertion that both
lated alcohols are converted to carbeniura ion in 70% HC10.
.
4
I, ' bis complexes dissociate in the acid, no Lncreas in
absorption Intensity is detectable. Thus, both coordinated alcohols
isl have br-ea ionized. Of course:, this contention could be definitively
established by titrating a weighed sample of a bis complex vs. the
corresponding mono complex, for which twice the quantity of water would
b< required to reconvert the bis complexed ions to the neutral precursor
i £ both coordinated alcohols had been ionized initially.
It is also pointed out that the values for the thermodynamic data
Ln Table V correspond to the reverse of the titration equilibria
(equations (2), (3), and {4}) as written. Hence, an examination of
these data in relation to the relative ease of carbenium ion foi'mation
affords a measure of the stability (as opposed to instability) of that
ion. In writing these equilibria it was tempting to consider the equa-
l Lons which obtain upon carbenium ion generation; viz., for a free
alcohol ion:
H+py^0H + 2 HC10
4-> H
+pyL
++ H
++ CIO" {13}
That is, this statement illustrates the net chemical change which occurs
upon dissolving a protonated pyridylmethanol in 70% HC10. and obviously
corresponds to carbenium ion formation. (It is not germane to consider
oton source required for the initial protonat.ion of the pyridine
' ,;.) It must be emphasized, however, that the carbenium ion stability
i from the aqueous titrations which are correctly
Lbed by equations {2}, {3}, and {4}: and it is for this reason
I this distinction has been mail
103
Finally, a comparison of the stability data for the mono complexes
to the data lor the corresponding bis complexes proves worthwhile. Those
data suggest thai related mono — bis complexed carbeniurn ions are of
practically identical thermodynamic stability. In the absence of. charge
effects this would have been an expected result. However, based upon
the consequence of like-charge repulsions on the stabilities of related
free 2-pyridyL vs_. 4-pyridylcarbenium ions, it seems reasonable that a
given bis complexed ion would be somewhat less stable than the correspond-
in; mono complexed ion. Apparentlystherefore, the degree of charge
separation in the bis complex ''di''-carbenium ion is sufficiently great
that there is r.o residual ion destabilizing effect exerted mutually
upon one another by the two coordinated ions in the complex. This
stability similarity is also reflected by a virtual superimposability
of the dilution curves for a related pair of these complexed carbeniurn
ions. Thus, the stabilizing influence exerted by the metal center on
the stability of such 4-pyridyl complexed ions appears to be independent
of none vs . bis coordination. Although, it may be the case that de-
stabilization arising through charge repulsions in the bis ions is
counterbalanced by enhanced stabilization through solvation in the highly
polar HC10, — H„0 solvent system. This possibility is suggested as a
consequence of the considerably greater solubility of the bis vs. the
mono comulexes in HC10, — K,0. Additional studies are required to test4 <c
' confirm the validity of this explanation.
The VrobJj-im^f_^rbenjum Ion Rea rrangement in HCIO 4 — H9O. Initial
attempts to gather stability information for the carbeniurn ions derived
fr. the free and complexed urisubstituted phenyl pyridylmethanols .led
104
'
' ual and unanticipated results. For instance, it was dis-
red Chat the dilution curves for the HC10 , — Ho titrations of these
carbenium ions could not be treated by the "1/2" Beer's law line extra-
polation technique in order to get the corresponding stability constant
data! That is, the second portion of the dilution curves (cf., Figures
10-12) obtained for any of these particular ions never fell off suffi-
i i ntly to intersect with the "1/2" Beer's law line. In other words,
the intensities of the principal electronic absorptions of these ionic
species were found to grow steadily on standing (until the intensities
had magnified by a factor of ca. 1.5 of original). In relation to this
19unanticipated result was the fact that the '" F nmr signal initially ex-
hibited by HC10. — HO solutions of these ions was found to disappear
with tinie (ca.. within 1-2 hours to 2k hours depending upon the original
concentration of the carbenium ion under investigation). The nmr result
suggested therefore that the para- fluorine phenyl ring substituent was
discharged kinetically. Shutske's results (68) verified this contention.
Shutske studied the reduction of various £luorospiro(isobenzofuran-
piperidine)s in 97% formic acid and discovered that on standing fluorine
was lost as a phenyl ring substituent in the parent compound. This
result prompted Shutske to repeat the investigations of Dayal e_t_ a]_. (69)
on the reduction of 4-fluorophenyldiphenylmethanol in 97% formic acid —
sodium formate solution. Shutske recognized the similarity of this
estigation to the fluorospiro study with respect to the anticipated
ction of the para-fluorine substituent. Dayal _et a 1 . had reported
isolation of only 4-fluo vldiphenylmothone as the reduction
." uct. Shutske, however, found that although this alkane was the
u Lpa] reduction product, 4-hydroxydiphenylmethane was also obtain
10!
In a yield of 13%. Consequently, the loss of the para-fluorine sub-
stituent from the unsubstituted phenyl pyridylmet ha 'in 1IC10, — H_0
paralleled Shutske's results. And. in retrospect, this result indicates
: it the ability of a 4-fluorophenyl group to participate in conjugational
interaction with the exocyclic carheniura ion carbon via it—electron
release Ls more substantial than is that of unsubstituted phenyl. This
is, in fact, borne out by the results of various studies on the thermo-
dynamic stabilities of trityl-type carbenium ions which also contain
4-fluorophenyl rings (18,19,70,71), etc. Hence, the degree of positive
charge which migrates by resonance into the 4-fluorophenyl ring in these
particular carbenium ions is significantly greater than for either of
the ion systems containing 4-methylphenyl, or 4-methoxyphenyl substituted
rings. Thus, in these latter cases the fluorine nucleus was not lost
during studies on carbenium ions in HC10. — HO.
The "Deno" titration stability data (Table V) for the unsubstituted
phenyl pyridylcarbenium ions were obtained in the following fashion.
The initial portion of the dilution was plotted as usual (viz . absorbance
vs . D.F.) to yield the Beer's law dependent region followed by the
normal smooth curve deviation from the extrapolated Beer's law straight
line. It can be seen by inspection of the dilution curves (Figures 10-
12) that the plot is then again linear over a substantial region of
the titration until a smooth curve upswing of the plot is reached which
thereby establishes that the titration is virtually complete. That is,
i e curves are very similar to those got from ordinary acid — base
titrations, with the difference that the nertinent carbenium ion equi-
i ria persist over a rather substantial concentration range of tb
Lzing medium (ca. 10 wt 7. HC10 , ) , resulting in a titration curve
Lth consid tbly less drastic chang i Ln ' than are customarily
>ited by acid — base titration curves. So, since the dilution curves
carbenium ions (i.e. , the ions which rearrange in HC10. — H, 0)
I<.t curve upswing prior to intersection with the "1/2" Beer's law
, the so-called second straight line portion of the curve (thai
portion between curve dox^nswing and curve upswing) was extrapolated
i Intersect with the "J/2" Beer's law line. This simple technique
Llitated the isolation of the D.F. value for which [R-OH] = [R+
]
,
and thereby allowed the determination of K . for the unsubstituted
phenyl carbenium ion in question. The reliability of tiie data ob-
tained by this method was tested by precalculating appropriate v;t %*s
HC10, required to yield D.F. samp Les whose carbenium ion absorbance
would tentatively lie on the extrapolated dilution curve in the region
oi interest. The D.F. samples tested in this fashion gave absorbances
which coincided very well with absorbances predicted from the extran-
ted plot. Hence, the stability data so obtained are considered
19to be very satisfactory. It will also be shown that F nmr chemical
shift data, in conjunction with attendant free energy relationships,
serve to establish that these data are acceptable, vide infra. (Not-:
Certain experiments were performed in attempts to determine the nature
of the species which resulted upon fluorine atom ejection. A brief
account of this work is provided in the Appendix.) Furthermore, it
[ally of interest to study the stability of such fluorophen
pyridyL nium ions as a function oF the rate of discharge of the
Luorine substituent. Indeed, these experiments should be relatively
straightforward to monitor via F nmr, and could provide additional
illation concerning the carl ion stabilizing influence as exerted
is coordinated metals.
L07
19F Chemical Shii I Carbenium Ion Si ! lv. To i , the results
19oi previous L' nmr investigations have. demonstrated that a para-sub-
Stituted fluorine nucleus on an aryl ring is a highly sensitive probe in
regard to the stability of carbenium ions which are conjugated with
the fluorine nucleus. Hence, further discussion respecting thy appro-
priate nature of this nmr technique shall not be presented, excepting,
19of course, as it pertains to the F nmr data obtained during the course
of this vrork.
19The use of F nmr chemical shifts as a criterion for the stability
of coordinated carbenium ions is, in itself, a novelty. Past studies
19which have employed F nmr measurements for the purposes of elucidating
the stability and structure of coordination compounds have normally
incorporated fluorine into the system as a ligand itself, i.e ., fluoride
ion (see, for instance, the nmr studies by Dixon and McFarland (72),
and references to related work cited therein). Also, fluorine in this
capacity has been used for nmr investigations as a substituent on a
ligand moiety bound directly to the metal center such as in the contro-
versial studies by Parshall (73,74). This author proportioned the
19measured F chemical shift of 3- and 4-f luorophenyl groups coordinated
to platinum(II) to the trans-directing abilities of ligands in the
complex located trans to the fluorophenyl groups. In any event, it is
believed that the current work embodies the first attempts to correlate
19F chemical shifts with the thermodynamic stabilities of metal-complexed
trityl—type carbenium ions.
19The ' F nmr spectra which were recorded for the free and complexed
pyrldylmethanol — carbenium ions are uncomplicated and highly similar.
I' absorption pattern was characterized by the presence of a
tuplel fluorine signal which is predicted for H— F coupling in
these systems by the relationship:
Number of peaks - (2 I-n, + 1) (2 l„n + 1) • • • (2 I n + 1)i i 2 2 n n
{14}
•re T - I - 1/2 for hydrogen, and n, = n« = 2 for the two pairs of
hydrogens situated ortho and raeta respectively, to the single fluorine
nucleus present in each species investigated. (The basic form of this
equation is given in (75).) The center peak of the 9-fold absorption
expectedly the most intense. The only difference of consequence
resulting from a comparison of all the spectra lies in the relative
positions of the signals for each species. For simplicity, the center
19of the 9-fold resonance pattern is taken as the F absorption position.
Hence, the position of the fluorine signal, recorded relative to an
appropriate reference standard, provides an indication of the stability
of the carbenium ion under investigation relative to the degree of
ir-electron delocalization present within the structural framework of
each, carbenium ion species. The reference standards employed were ex-
ternal CFC1_ and external trifluoroacetic acid (TFA) . The use of two
standards afforded a double check on the reliability of the obtained
shift data. External referencing procedures were, used for all carbenium
ion spectra owing principally to the extremely reactive nature of the
and of the HC10, — li?
solvent. Consequently, the absorption position
'i standard might have been appreciably affected by inter-
i Ions with solute, or solvent. Nonreactive, nonpolar materials, ob-
Ly could not be si i i tctorily employed as internal references
because of th< ; miscibility with HC10, -- II 0. Referencing against
i tie same externa] standards also facilitated the comparison of chemical
shift data obtained for the unionized ligands and complexes in acetone
to the related data for the corresponding carbenium ions in HC.10, — H.;0.
Bulk suseeptib.u i ty contributions by the solvent to carbenium ion chemical
shifts were expected to be unimportant in that each, ion spectrum was
recorded in 70% HC10, . Hence, solvent chemical shift susceptibility
contributions were leveled for each nmr measurement.
19The shift data resulting from these F nmr investigations are
1 19given in Table VI, pp. 110-111. Since H — F coupling is of no
19particular concern to this work, examples of the F spectra are not
given.) The following statements in relation to the relevant trends
established by these data are in order. A comparison of 6 values for
the alcohols in acetone, 1 K HC10, and 70% HC10, , systematically shows
the effect of proceeding from an aprotic solvent, to a protonating
solvent, to a solvent capable of carbenium ion generation. Consequently,
3 small downfield shift of the resonance for a given alcohol is detected
upon protonation (i.e
.
, acetone vs . 1 M HC10, as solvent) ; whereas
trans formation to carbenium ion is accompanied by a relatively substantial
19downfield shift in the F signal. This same trend is exhibited by the
complexes but, of course, no 1 M HC10, spectrum was recorded for these
materials as they are not soluble at this acid concentration. Hence,
as expected, conversion to carbenium ion results in appreciable resonance
dc Li calization of positive charge throughout the conjugated molecular
framework. In proceeding from R = -phenyl to R = -4-methoxyphenyl for
any related series of carbenium ions an appreciable upfield shift of
] signal is observed. This reflects the substantial capacity of
I
-J-
1
1
ucd
a «CD 4-1
>
r/1
</,
01
J
H>CD
i—
!
/-%
«
o!rH
iuIT. CJ
co
U
o
HC_>
o<oHu
SI
co!-(
uCJ
Ma*
3o
•Hd<u
cd
T3IT)
co U0
"O
Occ
CM
COro co
COro
CMrHu
oij
112
phenyl substituent relative to hydrogen, • • • and in turn
relative to a para-fluorine substituent, to remove po charge from
ocyclic carbon via conjugational interaction. A comparison of 6
for the 2-pyridyl vs. the corresponding 4-pyridy l.carbenium ions shows
the destabilizing effect of like-charge repulsions on carbenium ion
stability to be more significant in the 2-pyridylcarbenium ions. Hence,
19.the F signal for a given 2-pyridylcarbenium ion is downfield relative
to its more stable 4-pyridylcarbenium ion congener. This is in keeping
with the thermodynamic data which have been considered previously.
The consequence of coordination on a given F resonance is reflected
by a substantial increase in 6 found in proceeding from a free to the
corresponding complexed ion(s). This trend also parallels related
thermodynamic data obtained from the titrimetric investigations and
again indicates that complexation stabilizes unsubstituted phenyl-
carbenium ions to the greatest extent.
19For the data reported, it can be seen that the " F signals for
a related mono- and bis-complexed carbenium ion pair are virtually
19Ldentical. Since two F resonances were not initially detectable for
the bis complex ions, it is clear that both coordinated alcohols are
converted to carbenium ion in 70% HC10,; and again, a related mono-
b Ls-complexed carbenium ion pair are predicted to be of essentially
equal thermodynamic stability. Here, it is conveniently pointed out
19that when Lhe 70% HC10, F nmr samples of the mono and bis complexes
(of R = -4-methylphenyl, for example) were allowed to stand, eventually
two lignj Ls Lop d, 0i of the signals coincided with the original
m d carb< i sorption, whereas the newly developed signal
with the free carbeniui Lon resonance. These data verify tl
i L3
n i :t nature of the coordinate, bone! between palladium and the pyridii
ring in 70% ILC10 as the second signal was not detectable within I
fi st few hours of sample preparation. Furthermore., for the mono-
ed ion, the intensity of the free ion signal was not equal
(approximately) in magnitude to that of the complexed ion until the
iple had stood ca. 24 hours. Again, this demonstrates the stability
of the coordinate bond.
19The '" F nmr shift data reported for the carbenium ions derived
from the free and complexed unsubstituted phenyl pyridylmethanols are
considered to be reliable for the following reasons. Principally, the
19detection of a F signal in these systems for a freshly prepared sample
essentially guaranteed that the nmr measurement was made on the species
.19of interest. This follows since the originally detectable F signal
disappears only after rearrangement of these carbenium ions takes
place. Free energy correlations (considered i.n the forthcoming section)
also served to corroborate the credibility of the nmr data; and, as
well, instated the use of appropriate F chemical shifts for the deter-
mination of the thermodynamic stability of 2-pyridylphenyl-4-fluoro-
phenylcarbenium ion (recall that the stability of this carbenium ion
species was not ti trimetrically measurable). As this particular
carbenium ion was allowed to stand, several poorly resolved signals
1 9loped in the F spectrum. An attempt was made to fingerprint
19these absorptions _via. comparison with the F nmr spectrum of 1:5 48%
70% HC10, . The F signal for HF in 96% H o S0. reportedly exists
I -116.9 ppm relative to external TFA (76), hut no similar absorption
wa ' detectable in the HF — HCIO^ spectrum. Presumably, the HF reacted
the glass nmr tube to produce silicon fluorides or fluorosilicates.
: r side products were expected for the remaining unsubstituted
Lcarbeniura ion samples where HF was also the initially anticipated
taining rearrangement product. A comparison of all un-
itituted phenylcarbenium ion nmr spectra, hov/ever, yielded little
correspondence of resonance positions for the signals which developed
upi i rearrangement. (For the concerned reader: the 2-pyridylphenyl-
4-f luoropheny Icarbenium ion spectrum degenerated upon standing into two
ill-resolved signals at ^a. 52.3 and 63.9 ppm relative to external TFA;
and the HF — HCiO, spectrum exhibited four principal absorptions, at
52.6 (very broad), 63.1, 64.3, and 66.6 ppm, respectively, relative
to external TFA. And, the intensity of the signal at 64.3 ppm grew
rapidly during scanning of the sample.)
One final set of nmr experiments was performed in an attempt to
19relate ca rhenium ion — carbinol precursor equilibria to associated F
chemical shift data. The wt % HCIO, datum for 50% ionization of a given
rhenium ion species was obtained from the appropriate dilution curve
19— D.F. plot. A " F spectrum of a carbenium ion system in this wt A HCIO,
was expected to exhibit either two principal signals corresponding to
loci and alcohol, respectively, or one signal arising from rapid solution
equilibration of ion and alcohol (see p. 101). To carry out this in-
vestigation a solution of 4-pyridyl-4-methoxyphenyl-4-fluorophenyl-
hanol was prepared using 53.8% HCIO^ . This is the so-called 50%
acid for this carbenium ion precursor. The nmr spectrum of
this solution, how sver, exhibited only a single resonance which corre-
i qI with the posil Lon for the " F signal of this ion in 70% HCIO,.
i, no additional resonance was detect ,[>! for this system until
115
the acid was diluted to _ca. 46%. At this acid concentration a second
mance finally developed which corresponded to the signal exhibited
by the protonated pyridylmethanol precursor. These signals became of
approximately equal intensity at _cji. 44.5% acid, and at _ca. 43% acid
t:ie carbenium ion resonance disappeared entirely, This result is
unusual for the following reasons. First, the dilution curve — D.F.
data for this carbenium ion indicate that the carbenium ion — protonated
alcohol equilibrium is measurably significant over a minimum range of
8-10% acid (this is, in fact, the case for each of the titratable
carbenium ion species, see p. 105), whereas the nmr result suggests
that this equilibrium is measurable only over a rather limited range
of ca. 3%. Secondly, the nmr result indicates that this particular
carbenium ion is essentially 50% reconverted to protonated alcohol
precursor at 44-45% HC10, which is much less than the 53.8% value pre-
dicted from the titrimetric investigation. Hence, the nmr measurement
suggests that a considerably less concentrated ionizing medium is
capable of producing substantial conversion of protonated pyridylmethanol
to carbenium ion, and, as yet, this result is unexplainable. It should
be noted, however, that there is a significant difference in concentra-
tion required for carbenium ion nmr investigations compared co concentra-
tions necessary for carbenium ion electronic spectral investigations.
That is, in order to obtain acceptable nmr spectra for materials of
relatively high molecular weight which are dilute in the absorbing
nucleus (viss . , one fluorine atom per pyridylmethanol of M.W. ca. 280-
-?300 amu), it is necessary to maintain solute concentrations at 10 II,
19or 1 Lgher, in order to obtain detectable "F resonances, even with
currently available Fourier transform techniques. In fact, the
] 1
6
phenyl subsl I tuted carbenium ion was visually detectable in
HC10, by virtue of the characteristic solution color of the ion,
even though no nmr resonance could be obtained. Thus, concentration
u Lrer.ients for nmr studies are considerably greater than for elec-
tronic spectral measurements on pyridylcarbenium ions. Hence, the two
experimental methods cannot be cross-checked quantitatively owing to
the grossly different concentration ranges necessary for each measure-
ment. (Note: The same nmr dilution experiment was performed on the
bis complex o£ 4-pyridyl-4-methoxyphenyt4-fluorophenylmethanol (carbenium
ion}. However, this investigation proved futile as dilution induced the
precipitation of the complex.)
Although additional studies are required to resolve the problems
encountered in the so-called nmr dilution experiments, a potentially
fruitful study is suggested concerning the carbenium ions which eject
fluorine in HC.10. — H„0. That is, presume it is possible to fix a known4 2
concentration ratio of a given unsubstituted phenylpyridylmethanol and
corresponding carbenium ion via appropriate manipulation of HC10, con-
centration. This should establish two simultaneous equilibria: a) al--
,o\ interconversion with carbenium ion; and, b) carbenium ion inter-
conversion with the species resulting after fluorine atom loss. This
19investigation would obviously be monitorable by F nmr, and perhaps
new information concerning the stability of such types of carbenium
ions would be afforded from kinetic, as well as from thermodynamic
tndpoints.
iree Energy (LFE) Correla tions and Carbenium T on Stability. A
bei of analytical free energy correlations have been successfully
117
Lied to the thermodynamic data obtained for the pyridylcarbenii
[ ins considered in this work. These correlations are presented graphi-
cally (where appropriate) in this section and discussed principally in
terms of the relative stabilities of the various carbenium ion species.
A significant outgrowth of these considerations is the indication that
coordination, which has been shown to stabilize a given carbenium ion
relative to the protonated ion, destabilizes the ion relative to the
unprotonated , uncoordinated ion (vide infra) . This implies, rather ob-
viously, that the localized removal of "sigma" electron density occurring
upon coordination to the palladium containing moiety cannot be separated
from a concomitant reduction of ir-electron density in the pyridine ring.
Consequently, it appears that complexation (for the cases considered
herein) does in face destabilize the pyridylcarbenium ions.
Various studies (see, for instance, (12), (19), (77), (78), and
(79)) have demonstrated the suitable application of the famous Hammett
equation (80), log K = log K° + pa, towards the correlation of thermo-
dynamic parameters for arylcarbenium ions. Therefore it. was reasonable
to analyze the thermodynamic data obtained for the pyridylcarbenium
ions by similar methods.
The plots which are presented in Figures 13, 14, and 15, p. 118,
established the following relationships to be linearly dependent:
(i) Hammett "exalted" Za + constants vs. AG° of carbenium ionp-r —formation. (The. a + values employed were 0.00 for para-H; -0.07 for
para-F; -0.31 for para-CH-; and -0.78 for para-OCH (31). Curves 1, 2,
and 3, are for protonated 2-pyridyl, protonated 4-pyridyl, and bis-
palladium(II) complexed 4-pyridylcarbenium ions respectively.)
0.8_
0.6
, 0.4-
0.2-
. -|—
10.00
for 1
119
19(ii) Carbenium ion F nmr chemical shifts recorded relative to
i :ternal CFCI3 vs . related values of LG° . (Curves 4, 5, and 6, corre-
p id to the same carbenium ion sequence as in (i) above. The point
on carve 4 designated by the * is extrapolated to the abscissa to yield
AG" for the formation of unsubstituted phenyl-2-pyridylcarbeniun ion.)
19(iii) Hammett "exalted" Zo < constants vs. carbenium ion " F
P+ —nmr chemical shifts relative to external CFClo- (Curves 7, 8, and 9,
correspond to the same carbenium ion sequence as in (i) above.) The
ensuing discussion is in respect to these relationships. The appropriate
application of Hammett a + constants (rather than the usual a constants)
in free energy correlation analyses for arylcarbenium ions upon varia-
tion of electron donating, para-phenyl substituents has been established
principally through the work of Deno and Evans (78), and Broxvni and
Okaraoto (82). Therefore it was expected that these parameters were
found to correlate exceptionally well with the pyr idylcarbenium ion
stability data. Admittedly, it is seen that in each case the curves
have been drawn employing only three data points. None the less, the
degree of linearity obtained for all correlations considered is aston-
ishingly good. And, although in certain instances nonlinear free energy
cox-relations may prove to be interesting and important, a linear correla-
tion is the required sine qua non for the immediate and direct prediction
of interrelated but otherwise "unknown" thermodynamic information.
Indeed, it is on this basis that the stability constant data reported
for 2-pyriuylpheny] -4-f luorophenylcarbenium ion (Table V, pp. 95-96)
have been obtained and deemed reliable. That is, plots 2 and 3 in
Figure 13 served to establish the linear interdependency of T,a + and
for carbenium ion formation for the various 4-pyridyl species.
Similarly, plots 7, 8, and 9, have demonstrated the linear relation-
ship of Eo .j and 6 for each carbenium ion speci> estigated. Hence,
a-s shown, Che construction of plot 4 in Figure 14 using the nmr and
data tor the 2-pyridyl-4-methylphenyl, and 2-pyridyl-4-methox> phenyl
protonated carbenium ions, allowed the determination of AG° for the
corresponding unsubstituted phenyl ion. (Note: For these analyses the
data recorded relative to external CFC1.-. were used exclusively
owing to the greater significant figure accuracy obtained for carbenium
ton chemical shifts measured relative to this standard. Viz., as in-
dicated by che dnta contained in Table VI, pp. 110-111, the positions
of the sample and reference signals wi th external TFA as standard were
r?latively close. Consequently, only two significant figure reliability
could be obtainsd for the nmr data measured relative to external TFA.
I'rcheless, to check these LFE analyses, similar plots were drawn
using the TFA. referenced nmr data, and these plots also were linear.)
Two additional LFE plots of interest are illustrated in Figures
16 and 17, p. 121. The linearity of these plots suggests a proportionate
change in carbenium ion stability throughout the related 4-pyridyl vs .
2-pyridyl (Figure 16), and 4-pyridyl vs. bis-complexed 4-pyridyl (Figure
17), series of ions. The slope (= 0.829) of the line in Figure 16 re-
flects the lower stability of a given 2-pyridylcarbeniura ion compared to
its 4-pyridyl congener. Extrapolation of this curve to the intercept,
however, yields an inexplicable result from which it is implied that
for relatively stable pyridylcarbenium ions (i.e. , A0C very small, or
.tive) the 2-pyridyl ions are ultimately the more stable. Consequently,
it appears that extrapolation of this curve to a considerable extent
I the data points is not warranted. The slope (= 1.04) of Figure 17
123
50.00
15.00 ^
10 . 00 _.
v.C.°
(fecal)v mol '
5.00-
0.00
0.00 10.00ub ^kcax/mol)
20.00 24.00
Fig. 16. AC° (uncoordinated 4-pyridylcarbenium ions) vs. AG° (uncoordi-nated 2-pyridylcarbenium ions)
.
.kcal,
^mol }
20.001
15.00H
10.00-
5.00-
o.oo-
0.00
slope = 1.04
intercept = 1.60
10.00AG° (kcal/mol)
20.00 24.00
Fig. 17. AG° (uncoordinated 4-pyridylcarbenium ions) vs. AG° (bis-com-plexed 'r-pyr idyl carbenium ions).
122
roborates the c< ion previously a I upon that comp] : on
stabilizes th Least stable 4-pyridylcarbeniura ion to the greatest
extent. In ti> ; <; case the intercept value implies that a coordinated
pyridylcarbenium ion is inherently more stable than the corresponding
protonated icn. This is a reasonable result. Of course, identical
analyses of new, related data are required before more exacting con-
clusions can be. drawn. (Note: Several other linear plots are construct-
ive using the accumulated thermodynamic data owing to the established
linear iuterdependencies (as shown) of AG°, a + , and 5. These plots can
be synthesized as necessary for supplementary correlation studies.)
An examination of the slopes of the curves of Figures 13, 14, and
15 (1/p - -9.42 for curve 1; 1/p = -7.82 for curve 2; and i/p = -7.43
tor curve 3; these slopes are reported as 1/p for the comparisons which
% to be made) affords a simple method for comparing the stabilities
of i he pyridylcarbenium ions to related systems of arylcarbenium ions
vhich have been investigated previously via similar techniques. That
is, as pointed out by Freedman (64:1548), a plot of AC° vs. Y.c + for
a series of structurally and substitutional^ related arylcarbenium ions
yields a straight line whose slope reflects the degree of electronic
demand at the carbenium ion center. And, in these cases, a negative
e indicates that the presence of electron releasing substituents
capable of interacting conjugationally with the carbenium ion positive.
charge facilitates the development of that charge. Thus, carbenium ion
stabilit Ls enhanced by substituents such as para-methyl and para-
i i :y relative to p_ara-II, etc. Also, the magnitude of the slope is
tive measure of the stability of the particular family of ions
consideration. For example, p for plots of AG + vs. T.o + is on
the order of -2.5 for a .series of malachite green carbenium ions, -4.5
for a series of tritylcarbenium ions (64:1749), and -8 for para-sub-
stituted diphenylmethylcarbenium ions (64:1550). Therefore the sta-
bilities of the pyridylcarbenium ions are again shown to be in the order:
Coordinated 4-pyridyl > protonated 4-pyridyl > protonaLed 2-pyridyl;
and furthermore, these, results imply that the 4-pyridyl ions are but
minimally more stable than diphenylmethylcarbenium ions, whereas the
2-pyridyl ions are appreciably less stable.
Finally, the use of certain, appropriate arguments permit a reason-
able comparison of the stability of the protonated and complexed 4-pyr-
idylcarbenium ions to the stability of such ions premised upon a non-
bound pyridine nitrogen. That is, although it is not possible, obviously,
to investigate an unprotonated free ligand pyridylcarbenium ion in
strongly acidic media, it is necessary to speculate on the stability of
this ion in order to estimate the net stabilizing effect exerted by
the metal containing moiety on the coordinated ion. This is done in
the following fashion. The contribution to the stability of a triaryl-
type carbenium ion by an unprotonated pyridine ring may be approximated
using the literature pKj,+ values for malachite green (MG) carbenium
ion (pK^-f. = 7 07) and 4-pyridine malachite green (4-pyMG) carbenium
ion (pK^j. - 5.66). These data have been taken from the compilation of
Nemcova et al . (83). (Note: Here T>K,+ corresponds to the equilibrium
reconversion of carbenium ion to precursor alcohol: R + 2 H„0 J R-OH
+ ELO ; and 4-pyMG is simply MG with the unsubstituted phenyl ring re-
placed with a 4-pyridyl ring.) An examination of these pKp+ values
indicates that exchange of a phenyl ring for a 4-pyridyl ring induces
a net carbenium ion destabilization of 1.41 pK units (1.92 kcal) in
124
Les of ions. This destabilization may he quantitatively
to strictly triphenyl-type carbenium ions (i.e. , those without
Lno Bubstituents) by comparing the slopes (p's) of the plots
of log K vt>. £0 + for the MG series of ions (p = -1.4) to that for the
triphenyl ions (p - -3.5), (84:101-102). That is, as pointed out by
Hine, phenyl ring substitution in the triphenyl ions induces a consider-
ably greater change in carbenium ion stability than does the identical
titution in the MG series of ions. Hence, the larger negative slope
oi this plot for the triphenyl ions reflects the greater sensitivity of
the stability of these ions to electronic effects. This consideration
it. illustrated through a simple analysis of the pK data given below.
These data are in respect to the general equilibrium described above.
(i) pKd+ of triphenylcarbenium ion = -6.63; and pK_}_ of para-
t'ltrotriphenylcarbenium ion = -9.15 (10).
(i l) pkp+ of MG carbenium ion = 7.07; and pK~o+ of para-nltroMG
c u-benium ion = 6.00 (64:1534). Thus, as expected, the destabilizing
effect exerted by a para-nitro substituent is much greater in the tri-
phenyl scries of carbenium ions (-2.52 pK units) than in the MG series
of carbenium ions (-1.07 pK units). And, the ratio of these desta-
ging contributions, 2.52/1.07 = 2.36, is certainly comparable to
that predicted from the slope ratio, 3.5/1.4 = 2.5. Now, in that pyr-
idine exhibits tendencies towards electrophilic aromatic substitution
Lch nre similar to nitrophenyl rings, and since both of these systems
illize trityl-type carbenium ions, it is reasonable to predict p]
unprotonated 4-pyridyldiphenylcarbenium ion to be on the order of
-6.82 - 2.5(1.41) = -10.34 = ?K,+. (Mote: For this approximation -6.82
. iue for plvp+ of triphenylcarbenium ion. This is the average of
125
the pK_+ values cited in Table V, pp. 95-96,. for this ionic species
in 70" HC10,.) A comparison of this value (-10,34) to the pK + values
which have been determined for the protonated (pKp+ = -13.47) and bis-
complexed (pKj.+ - 12.21) 4-pyridylphenyl-4-fluorophenylcarbenium ion,
strongly suggests that the unprotonated ion is appreciably more stable
Chan either the protonated (obviously) or the palladium complexed ion.
Moreover, the para-fluoro substituent would contribute additional sta-
bilization to this ion such that the actual pK^+ would be slightly
greater (less negative) than the speculative value of -10.34. This
further substantiates the validity of the hypothetical stability com-
parison.
Suitable use of the log K vs_. Za + plots given by Hine (84:101)
in conjunction with the results of Barker et_ al. (52) affords another
simple corroboration of this consideration. That is, examination of
the plot for the MG series of ions (Hine) reveals that for MG with pK^+
= 7.07, Zcr . = -3.4; and for 4-pyMG with pK^ = 5.66, Za + = -2.5. Thus,
with a + = -1.7 for para-N(CH„)„, a hypothetical value of a = 0.9 is
predicted for a 4-pyridyl ring, thereby demonstrating the electron
withdrawing capacity of a pyridine ring taken as an arylcarbenium ion
substituent. This value is comparable Lo a =0.78 for a nitro group
as a para-phenyl substituent; and the ratio 0.9/0.78 = 1.2 is in good
agreement with the ratio 1.41/1.07 =1.3 which is obtained from the
respective energetic destabilizations of a 4-pyridyl ring and a 4-nitro-
phenyl ring in MG carbenium ion. Similarly, the linear correlation of
(kK) with Hammett a substituent constants discovered bv Barker
and coworkers for a series of phenyl substituted MG carbenium ions
.ion of a of C2. 1.0 - 1,1 for the 4—pyridyl ring in
relation is made b;y using the value of 15.4 kK re-
al: A for 4-pyIIG (83). Furthermore, use of the secondmax
plot presented by Hine of log K vs . L"o for triphenylcarbenium
ions, and the provisional value of ca. -10.3 for pK + of unprotonated
''t-^yridylphenylcarbenium ion, allows a value of o of ca. 1 to be ob-
tained for the 4-pyridyl ring. Thus , the hypothetical a values which
are predicted for a 4-pyridyl ring treated as a substituent in arylcar-
benium ions are found to be in very good agreement. Therefore, on this
basis, it is reasonable to conclude that an unprotonated pyridine ring
is of approximately the same electron v/ithdrax^ing capacity as a nitro-
pneuyl ring, but, as suggested previously, is not as electron withdrawing
as the palladium coruplexed pyridine ring.
A last extension of these considerations can be made employing
trie results of Atkinson e_t al. (85). These workers satisfactorily
demonstrated the existence of a linear correlation between the frequen-
cies (v, kK) of the longest wavelength electronic absorptions and Taft
o"-'c polar substituent constants for a family of diphenylmethylcarbenium
+ions (<f;„-C-X) generated in 96% ILSO,. The group X is the substituent
which is varied in this series of ions. Here, a* reflects the electron
releasing or withdrawing capacity of X depending upon the exertion of
Lar effects by X. Since the 4-pyridylcarbenium ions are of comparable
stability to the diphenyl ions (p. 122) and have exhibited about the
same sensitivity to changes in electronic environment at the carbenium
it ,i center an these ions (recall the similai p values for plots of AG°
f- for these ions), it is reasonable to estimate values of o* for
protonated and coordinated pyridine rings taken as polar substituents.
127
[n fact sthis treatment may indeed be more logical than the. a evaluation
ag to the polar character produced in the pyridine rin^ by protona-
tion (and, apparently, by coordination as well). Inasmuch as the elec-
tronic spectral investigations have indicated that the absorption at
,\ reflects principally electronic interactions between the carbeniummax
ion centex and the phenyl rings, it is necessary to consider the fre-
quencies at A for the protonated and bis-palladium complexed unsub-
stituted 4-pyridyldiphenylcarbenium ion. The frequencies are 21.0 kK
and 21.4 kK, respectively, for these ions in 96% H,.S0, (7). (Note:
The electronic spectra of these ions are not changed in 70% HC10, .)
Using the v — a* correlation plot (85) hypothetical values of a* of _c£.
1.0 for protonated pyridine, and a* of ca. 0.85 for palladium-complexed
pyridine are obtained. These a* values indicate rather substantial
electron withdrawing capacities for these pyridine moieties. Again,
based upon the comparable stabilities expected for 4-nitrotriphauyl-
earbenium ion and unprotonated 4-pyridyldiphenylmethylcarbenium ion,
a speculative value of a* can be estimated for the unprotonated pyridyl
ring. That is, since v at X for the nitro ion is 22.0 kK, whichmax
correlates with a* of ca. 0.6, a value of a* of ca . 0.6—0.7 is pre-
dicted for the unprotonated pyridyl ring. Of course, this speculation
rests on the assumption that these two ring systems influence the fre-
quency at X to approximately equal extents. This contention ismax
supported by the fact that these rin.-
; systems have been shown to be of
comparable electron withdrawing capacity. Therefore, as indicated
previously, the palladium(II) containing moiety employed in this work
as the coordinating agent, appears to contribute an overall thermo-
dynamic destabilization to trityl-type pyridyl carbenium ions.
IN CONCLUSION
The. experimental bases which have been established for the investi-
gation of heteroaromatic carbenium ions (6, 7, and 8), have been re-
inforced and expand ed by this work. The achievements which are felt to
be o£ principal significance are. the following:
(i) The preparation of the mixed "mono" alcohol complexes (denoted
in Che text as [Pd(II) (pyLOH) (L )C1„]) appears to be the first success-
ful application of a synthetic method generally suitable for the selec-
tive incorporation of strong ir-acid ligands into the coordination sphere
c^ palladium(II) , to include pyridine-type donors. Consequently, it is
now expected that pyridine containing arylcarbenium ion precursors can
be introduced as ligands ad libitum into palladium(II) , as well as-
certain other, metal centers. Therefore, many additional, related
studies on complexed carbenium ions have been made feasible by the
development of this synthetic technique.
(ii) The thermodynamic stabilities determined for the various
caibeniuni ion species considered herein provide strong evidence to in-
dicate that charge repulsion effects tend to destabilize the 2-pyridyl
ions to a considerable degree. On the other hand, the very similar
L ties found for related "mono" and "bis" complexed carbenium ions
indicate bhat charge repulsion effects in the doubly ionized "bis" ion
terns are no longer sufficient to destabilize these ions. So, it is
128
ntially of value to prepare several "bis" carbeniura ion complexes
Ltl the Interesting prospect of determining the sensitivity of the
Stability of a particular completed carbenium ion to intra-species
charge repulsion effects. Moreover, carbenium ion stabilization provided
by various metal-containing coordination centers can be considered in
respect to this possibility. That is, effects on carbenium ion stability
resulting from variation of the central metal species, variation of
the oxidation state of the metal species, variation of the counter ligands,
etc.;, can now be investigated with similar intent.
(iii) Again, the results obtained from the "Deno" titration
stability measurements, established to be reliable by appropriate LFE
analyses, have shown that this titrimetric technique is a suitable
method by which to evaluate the thermodynamic stabilities of protonated
and complexed pyridylcarbenium ions. Furthermore, successful electronic
spectral interpretations have provided experimental and theoretical
substance towards correlating arylcarbenium ion electronic absorptions
with the aromatic ring systems which comprise the carbenium ion entity.
Consequently, it has been made possible to assess tentatively the con-
tribution to coordinated carbenium ion stability exerted by various
metal containing moieties in terms of the relative degree of influence
upon the frequency of the so-called y-band electronic absorption as
measured for the different metal centers.
19(iv) The applicability of F nmr chemical shift measurements as
a criterion of the stability of pyridylcarbenium ions has been established.
That is to say, a fluorine nucleus para substituted on a phenyl ring
been shown to be a suitably sensitive probe for the purpose of
quantitatively estimating the stabilities of these types of carbenium
13(1
u is. A possibly fruitful extension of this work might well be provided
l 9by emeuts with the fluorine nucleus incorporated as a
phenyl substituent in the carbenium ion molecular framework. By
a comparison of the results of these nmr investigations with the nmr
dsf:t at hand, a separation of a- vs. 7i-eleetronic effects within the
py .' idyl carbenium ions is potentially achievable.
(v) The LFE correlations which have been applied to the. thermo-
19dynamic stability data, F nmr data, and the appropriate Hammetc-type
substituent constants, have proved to be exceedingly good even though
only three data points were used for each analysis. Obviously, the
preparation and investigation of additional pyridylcarbenium ion species
by varying para phenyl ring substituents, affords a simple means of
extension of the current work. The stability predictions made in
reference to unprotonated 4-pyridyldiphenylcarbenium ion could be pursued
experimentally via the syntheses of salts of the various pyridylcarbeniura
is, followed, of course, by appropriate measurements to determine the
stability of these ions. The preparation of stable salts of the coor-
dinated carbenium ions could well be stimulating and provocative owing
to the stringently anhydrous, nonbasic, conditions required in order
to sustain the existence of such materials. Consequently, it. could
prove difficult to develop a preparative route during which the metal
containing moiety would remain coordinated during the conversion of
oho! to carbenium ion salt.
To recapitulate then, the experimental studies on the various
,ienium ion species which have been carried out during this work have
i very successful. It is therefore proposed that the scope of
lamic investigations upon free and complexed heteroaromatic
benium ions can now be broadened considerably by extending these
to include related systems of arylcarbenium ions.
APPENDIX
1_. Specific Gravit y of Aqueous HCIO^ Determined as a Func t ion of
Hi HCIO4 . By employing the. specific gravity (n) — wt % HC10, data in
ible TV. p. 91, a computer fitted "least squares" estimate of p as
a function of wt % HC10, was obtained. This was carried out by assuming
the functional relation y(p) = f(x) (x = wt %) to be of the form:
2 3 4a + a,x + a„x + a„x + a,x {15}
for y vs . x which yields a smooth curve plot over the entire range of
experimental (x,y) values. Then, the values of the coefficients (a's)
for this polynomial are obtained from a least squares estimates analysis
by minimizing the sum of the squares of the deviations which result
im:
2 2(deviation) = [y(calcd from {15}) - y(exptl)] {16}
sucamad over all the experimental points. With the a's calculated,
equation {15} is:
y = 0.996357669170ex 00 -h (0. 606617134345ex - 02)x -
(0..171560538090ex - 04)x2
+ (0. 164401851371ex - 05)x3 -
(0.98l691970813ex - 08)x4
{17}
from v,hidi a complete series of p values as a function of wt % HC10,4
is obtained. (Note: it was found that the x and x terms may be
n '.Lected with no loss in data precision to 4 significant figures.)
le data were processed at intervals of 0.010 wt % spanning the range
]33
134
p = 0.996357 at wt Z HC10, = 0.000 (i.e. , pure water at ?.5'J ), top =
1. '37627 at wt % HC10, = 75.000. The p values were rounded to A
significant figures - or tne required D.F. calculations. The program
employed was: WAHG Series 700 V 1. General Library, Program 1063A/ST3
- "N order regression analysis". A general description of the method
is provided by Kuo (86)
.
2. Degree of Protonation of a Pyridylmethauol (pyLOH) in Aqueou s
HCIOa In reference to the assumption that a pyridylmethanol is of
approximately the same basicity as pyridine (p. 99), straightforward
acid-base equilibrium calculations may be employed showing that the
degree of pyridine ring protonation is virtually 100%:
(i) In connection with this assumption the following equilibria
can be written:
pyLOH + I12 t H
+pyL0H + 0H~ {18}
and
H+pyL0H + H
2t pyLOH + H
3
+{19}
-9(ii) Using {19} and K, ~ 2 . 3 x 10 ' for pyridine, the equilibrium
constant for {19} can be evaluated:
v = [pyLOH] [H-tO+
] [OH ] _ K| . [H+pyLOIf] [0H-] K^
K* :
9!
"""''~ / -, in" 6- -
2#3 x 10_9• 4.3 x 10
(iii) Now, with the conditions specified as [H.,0 ] = 1 and
[pyLOH]. . - 0.01 (p. 99), it follows that for
pyLOH + H3 t H pyLOH + H
2{20}
1 1 _ o o -m 5
\Z0j K f , 4.3 x 10 D1. 19 1
and. if [H+pyLOH] - x,
v - [H+pyL0H] x 5K
{20] [pyLOH] [H30+] (0.01 - x) (1 - x)
/J X 1U
ai d
(0.0.1 - x)(l - x) 0.01 - xsince 1 > > 0.01 > x
and for
_JL__ „ 105 = 10*
0.01 - xXU 10-7
x is on the order of ca. 0.0099999! Hence, within the limits of the
calculation, a given pyridylmethanol at ca. 0.01 M (initially) is
protonated to an extent of 100% in 1 M [HO J.
3-_ '
:
.' ,;.
Conducted upon the Carbenium Ion Species Foj
to Exhibit ;ement in 70% HC10& . As indicated by previous in-
which have employed acidic media to generate arylcarbeniun
CO I lining para-fluorophenyl substitueats, the fluorine atom
may eventually be displaced resulting in the formation of either an
alcohol (68) or a quinone (79). Consequently, loss of the para-
fluorine substituent from the pyridylphenyl--4-fluorophenylcarbenium
ions in the HCl^ — H„0 was not unusual since in these carbenium ion
systems the. para-f luoro substituent; is the predominant electron-
donating entity involved in conjugational interaction with the car-
benium ion center. Therefore positive charge build-up at the para-
position in the fluorine-containing ring ultimately results in nucleo-
ptiilic displacement (presumably by H-O solvent molecules) of the fluorine
substituent.
In recording the electronic spectra of these pyridylphenyl-4-
fluorophenylcarbenium ion species it was observed that the intensity
of the main absorption bands (the x and y bands) grew steadily upon
standing until the intensities had approximately doubled. Moreover,
it was observed that the positions of these bands remained essentially
unchanged during rearrangement of the uncoordinated ions, whereas the
mono-complexed [Pd(II) (L )Cl„pyL] carbenium ion exhibited an x-band
shift to 494 nm after standing a period of 12 days. (Recall that
A for this carbenium ion as the protonated species is 482 nm, and
for this particular complexed ion is 477 nm) . Thus, it is seenmax
"new" A (494 nm) is appreciably longer even than A formax ' max
ee (protonated) ion (482 nm) . Also, the bis-complex of this ion
Lbited a shift in A from 470 nm to 482 nm after standing a period
! 17
of 6 days; this species, however, exhibited no further red shift of
x-band absorption ar. did the corresponding mono-complexed ion
(above). These results are as yet not explicable.
In an attempt to estimate the relative stability of the rearranged
product, assuming it to be a new arylcarbenium ion, this species was
titrated via the usual method. This was carried out by preparing a
fresh sample of 4~pyridylphenyl-4-f luorophenylcarbenium ion in 70%
HClO^j and titrating immediately. The sample was then allowed to stand
ca . 24 hours during which time the intensity of the x-band absorption
had essentially doubled. This result indicated that although the orig-
inal carbenium ion had been substantially reconverted to protonated
alcohol precursor by titration, the protonated precursor was still
capable of rearranging to the new species thereby implying that the
new species was a more stable carbenium ion. After the intensity of
the x—band had become constant, indicating that rearrangement was
complete (under these conditions), the titration was repeated. This
second titration required approximately three times as much titrant
(water) in order to effect the same absorption fall-of? as compared to
the Initial titration. So, again, this species appeared to be a more
stable carbenium ion than the original pyridylphenyl-4-fluorophanyl
ion from which it was derived.
Finally, a sample of 4~pyridylphenyl-4-f luorophenylmethanol
(ca. 100 rag) was dissolved in 70% HC.10, (ca. 10 ml of acid) to completely
invert it to carbenium ion. Rearrangement was again monitored spec-
troscopically, and after it was apparent that no further changes in the
tronic spectrum of the sample were occurring, the acid was neutral-
ized by the careful addition of reagent Nil-. This resulted in the
138
of a mixture of a white crystalline solid and a rather gummy
Low solid. The mixture of solids was transferred to a filter and
wi :
; CO] Lous quantities of deionized water. The white solid
was readily removed by the washing, and the yellowish material remained
on i:he filter. The contents on the filter were, air dried to yield ca .
1!) mg of a yellow-orange amorphous solid. This material exhibited
decomposition to a brown-black oil above 140° and an ill analysis
revealed that this material was different than the original alcohol.
A visible spectrum of this material in 70% HC10, was identical to that
19previously obtained for the so-called rearrangement product. The. " F
nmr spectrum of this material in 70% HC10, was also recorded. Following
3000 scans with the Fourier transform synthesizer two weak signals
(^'. 37.7 ppm and 68.9 ppra upfield with respect to external TFA) were
detected. These signals did not; correspond to any of the nmr signals
which had been recorded relative to external TFA during previous
carbenium ion nmr analyses; and the very low intensities of these signals
indicat€:d the. absence of any appreciable concentration of fluorine in
the sample. Mass spectral cracking analyses of this material revealed
the presence of molecular ion fragments of nominal mass as high as 520.5
ami'.; and, at relatively low probe temperatures (200° or less) a 259 amu
peak was consistently observed. Interestingly, the 259 peak
corresponds to the mass of the quinone species which would result from
combined effects in the parent alcohol, of quinone formation at the
original fluorine site and hydroxy! loss at the exocyclic carbon. Care—
analyses of these spectra, however, suggested the presence of as
many as four new compounds with the distinct possibility of one or
more of these materials existing as a polymer. Rather obviously
139
refore, the nature of the material (s) produced upon fluorine atom
displacement from the pyridylphenyl-4-fluorophenylcarbenium ions in
I LO, H_0 remains to be ascertained,
BIBL10CRAPHY
1. S. P. McManuS and C. U. Pittiaan, Jr., in "Organic Reactive Inter-mediates," S. P. McManus, Ed., Academic Press, New York andLondon.. 1973, pp 200-201.
2. G. A. Olah, J. Am. Chem. Soc, 94_, 808 (L972).
'3. J, P. Wibaut, A. P. de Jonge, H. G. P. van der Voort, and P. Ph.
H. L. Otto, Rec. Trav. Chim. , 70, 1054 (J951).
4. H. A. Smith and C. W. Holley, J. Am. Chem. Soc, 80, 3714 (1958).
5. F. ft. Hartley, "The Chemistry of Platinum and Palladium," AppliedScience Publishers, Ltd.. Ripple Road, Barking, Essex, England,1973.
6. T. W. Richardson, Doctoral Dissertation, University of Florida,1971.
7. B. C. Bhat tacharyya and R. C. Stoufer, unpublished t-esults.
8. P. W. Went?., Doctoral Dissertation, University of Florida, 1972.
9. R. G. Turnbo, D. L. Sullivan, and R. Pettit, J. Am. Chem. Soc,86, 5630 (1964).
JO. N. C. De.no, J. J. Jaruzelski, and A. Schriosheiin, J. Am. Chem.
Soc, 77, 3044 (1955).
LI. K. C. Dene, J. E. Berkheimer, W. L. Evans, and H. J. Peterson,
.]. Am. Chem. Soc, 81, 2344 (1959).
12. R. ''• Taft and L. D. McKeever, J. Am. Chem. Soc, 87, 2489 (1965).
1.3. S. V. McKinley, J. W. Rakshys, Jr., A. E. Young, and II. K.
Freedman, J. Am. Choir,. Soc, 93, 4715 (1971).
R. W. ijiL, F. Prosser, L. Goodman, and G. T. Davis, J. Chem.
B, 380 (1963).
13. R. W. Taft, E. Price, 1. R. Fox, I. C. I.ewj K. K. Andersen, and
G. T. Davis, J. Am. Chem. Soc, 85, 3146 (1963).
M. J. S. Dewar and A. P. Harchand, J. Am. Chem. Soc, ^38, 3318
(1966)
.
140
141
17. R. G. Pews, Y. Tsuno, and R. ".,T
. Taft, J. Am. Chem. Soc, 89
.
2391 (196 7).
18. R. Fi.Jler.. C-S . Wang, M. A. McKinney, and F. N. Wilier, J. Am.Chem. Soc, 89, 1026 (1967).
19. I. I. Schuster, A. K. Colter, and R. J. Kurland, J. Am. Chem.Soc, 90, 4679 (1968).
20. A. Mendel, J. Organomet. Chen-., 6, 97 (1966).
21. F. J. McCarty, C. H. Tilford, and M. G. Van Campen, Jr., J. Am.Chem. Soc, J79, 472 (1957).
22. A. P. de. Jonge, H. J. den Ke.rtog, and J. P. Wibaut, Rec. Trav.Chim. , 70, 989 (1951).
23. F. B. LaForge, J. Am. Chem. Soc, 50, 2484 (1928).
24. F. G. Mann and D. Purdie, J. Chem. Soc, 873 (1936).
25. J, Chatt and L. M. Venafczi, J. Chem. Soc, 2351 (1957).
26. J. Chatt and D. M. P. Mingos, J. Chem. Soc A, 1770 (1969).
27. T. Boschi, B. Crociani, M. Nicolini, and U. Belluco, Inorg. Chim.Acta, 12, 39 (1975).
23. R. J. Goodfellow, P. L. Goggin, and D. A. Duddell, J. Chem. SocA, 504 (1968).
29. D. M. Adams, P. J, Chandler, and R. G. Churchill, J. Chem. SocA, 1272 (1967).
30. P. M. Henry and 0. W. Marks, Inorg. Chem., 10, 373 (1971).
31. D. W. A. Sharp and N. Sheppard, J. Chem. Soc, 674 (1957).
32 H. J. Dauben, Jr., L. R. Honnen, and K. M. Harmon, J. Org. Chem.,25, 1442 (1960) .
33. G. A. Olah, J. J. Svoboda, and J. A. Olah, Synthesis, 544 (1972).
34. T. M. Dunn, in "Modern Coordination Chemistry," J. Lewis and R. G.
Wllkins, Eds., Interscience. Publishers, Inc., New York, 1967,
p 281.
35. R. S. Mulliken, J. Chem. Phys . , ]_, 14 (1939).
;'>. N. C. Deno, P. T. Groves, and G. Saines, J. Am. Chem. Soc, 81 ,
5790 (1939).
17. C. S. Schoepfle and J. D. Ryan, J. Am. Chem. Soc. 54_, 3687 (1932)
L42
M. S. Ie nan and N". C. Deno, J. Am. Chem. Soc. , 73, 3644 (1951).
39. G. N. Lewis, T. T. Magel, and D. Lipkin, J. Am. Chem. Soc, 64,L774 (1942).
N. C. Deno, J. J. Jaruzelski, and A. Schriesheim, J. Org. Chem.19, 155 (1954).
41. G. Branch and II. Walba, J. Am. Chem. Soc, _7_6, 1564 (1954).
42. A. G. Evans, J. A. G. Jones, and G. 0. Osborne, J. Chem. Soc,3803 (1954).
43. R. Dehl, W. R. Vaughan, and R. S. Berry, J. Org. Chem., _24, 1616(1959) .
44. N. C. Deno, Progr. Phys. Org. Chem., 2, 178 (1964).
45. G. A. Olah, C. U. Pittman, Jr., R. Waack, and M. Doran, J. Am.Chem. Soc, 88, 1488 (1966).
*
46. A. Stre I.Lwleccr , Jr., "Molecular Orbital Theory for OrganicChemists," John Wiley and Pons, Inc., New York, London, andSydney, 1961.
47. R. Waack and M. A. Doran, J. Phys. Chem., 68, 1148 (1964).
48. R. Grinr.er and S. F. Mason, Trans. Faraday Soc, 60, 264 (1964).
49. G. N. Lewis and J. Bigaleisen, J. Am. Chem. Soc, 6_5, 2101 (1943)
50. Y. K. Mo, R. E. Linder, G. Earth, E. Bunnenberg, and C. Djerassi,J. Am. Chem. Soc, 96, 3309 (1974).
51. H. P. J. M. Dekkers and Ms. E. C. M. Kielman-Van Luyt, MolecPhys., 31, 1001 (1976).
52. C. C. Barker, M. [I. Bride, G. Hallas, and A. Stamp, J. Chem.Soc , 1285 (1961).
53. H. C. Longuet--Ui-gins, J. Chem. Ph\s., 18, 265 (1950).
54. A. C. Fopkinson and P. A. H. Wyatt, J. Chem. Soc B, 530 (1970).
55. B. G. Ramsey, J. Phys. Chem., 70, 611 (1966).
56. M. W. Jov.ch, M.S. Thesis, University of Florida, 1966.
57. M. V.'. Couch, Doctoral Dissertation, University of Florida, 1969.
58. II. J. Dauben, Jr. and J. D. Wilson, Chem. Cotumuns., 1629 (1.968).
59. G. A. Olah, C. A. Cupas, and M. B. Comisarow, J. Am. Chen. Soc,> (J 966).
60. G. N. Lewi : and M. Calvin, Chera. Revs., 25, 273 (J93 ;;).
61. A. K. Ci Lngton, M. J. Tait, and W. F. K. Wynne-Jones , Proc. Roy.
Soc. A, 286 , 235 (1965).
62. R. J. Gillespie, Endeavour, 32, 3 (1973).
63. C. A. Olah, Chem. Britain, 8, 231 (1972).
64. H. H. Preedman, in "Carboniura Ions," Vol. IV, G. A. Olah and P.
v. R. Schleyer, Eds., Wiley Interscience, New York, London,Sydney, and Toronto, 1973.
65. M. Herlem, Pare Appl. Chera., 49, 107 (1977).
66. 0. Redlich and G. C. Hood, Disc. Faraday Soc, 24, 87 (1957).
67. L. H. Brickwedde, J. Res, Nat. Bur. Stand., 42. 309 (1949).
68. G. M. Shutske, J. Org. Chera., 42, 374 (1977).
69. S. K. Dayal, S. Ehrenson, and R. W. Taft, J, Am. Chera, Soc,94, 9113 (1972).
70. R. Cigen and C. G. Ekstrom, Acta Chem. Scand., 17,2083 (1963).
71. E. D. Jenson and R. W. Taft, J. Am. Chem. Soc, 86, 116 (1964).
72 K. R. Dixon and J. J. MaFarland, Chem. Communs., 1274 (1972).
73. G. W. Parshall, J. Am. Chem. Soc, 86, 5367 (1964).
74. C. W. Parshall, J. Am. Chem. Soc, 88, 704 (1966).
75. D. J. Pasto and C. R. Johnson, "Organic Structure Determination,"
Prentice-Hall, Inc., Englewood Cliffs, N.J., 1969, p 182.
76. K. Baum and H. M. Nelson, J. Am. Chem. Soc, 88, 4459 (1966).
77. N. C. Deno and A. Schriesheim, J. Am. Chem. Soc, 11', 3051 (1955).
78. M. C. Deno and W. L. Evans, J. Am. Chem. Soc, _79, 5804 (1957).
79. S. V. Kulkarni, R. Schure, and R. Filler, J. Am. Chem. Soc,95, 1839 (1973).
80. L. P. Hammett, J. Am. Chem. Soc, _59, 96 (1937).
81. J. Shorter, "Correlation Analysis in Organic Chemistry,"
Clarendon Press, Oxford, 1973, p 14.
82. H. C. Brown and Y. Okamoto, J. Am. Chem. Soc, 80, 4979 (1958).
83. t. Nemc >va, M. Malat, and R. Zahradnik, Collect. Czech. Ch
Conununs., 34, 2880 (1969).
Lne, "Physical Organic C'aemistry," Second edition, McGraw-Hill,ED ; York, 1962.
85. A. Atkinson, A. C. Kopkinson, and E. Lee-Ruff, J. Chj.m. Soc.
,
PerkinTrans.il, 1854 (1972).
86. S. S. Kuo, "Numerical Methods and Computers," Addison-Kc-sley
Publishing Co., Reading, Mass., 1966, Chapter 11.
BIOGRAPHICAL SKETCH
James CharJ.es Horvath was born June 29, 1942, in Toledo, Ohio.
He was very fortunate to recognize that learning is a wonderful
experience which life forever offers. Who knows what may lie ahead'
I 5
I certify that I have read this study and that in my opinionit conforms to acceptable standards of scholarly presentation andis fully adequate, in scope and quality, as a dissertation for thedegree of Doctor of Philosophy.
uituUL-R. Carl Stoufer, ChairmanAssociate Professor of Chenistry
£yH -
1 certify that I have read this study and that in my opinionit conforms to acceptable standards of scholarly presentation andis fully adequate, in scope and quality, as a dissertation for thedegree of Doctor of Philosophy.
^LJflJL^klierle A. BattisteProfessor of Chemistry
I certi thai E li * . this study and that in my opiniononforms to acceptabJ • dards oi irl; pi ation and
Is fully adequate, in scope and quality, as a dissertation for thedegree of Doctor of Philosophy.
JJud^d^jJjamARichard D. DresdnerProfessor of Chemistry
I certify that I ha\e read this study and that in my opinion
it conforms to acceptable standards of scholarly presentation and
is fully adequate,, in scope and quality, as a dissertation for the
degree of Doctor of Philosophy.
George E. fltysch'.cewitsch
Professor of Chemistry
I certify that I have read this study and that in ray opinion
i : conforms to acceptable standards of scholarly presentation and
is fully adequate, in scope and quality, as a dissertation for the
degree of Doctor of Philosophy.
&*SJack R. SmiProfessor of Electrical Engineering
This dissertation was submitted to the Graduate Faculty of the Department
oi Chemistry in the College of Arts and Sciences and to the Graduate-
Council, and was • ;. i ted as partial fulJ Lllmenl hie re its
for the degree of Doctor of Philosophy.
M 1978
Dean, Graduate School
<>
AM2 3 8. 10.6 0.